Orca

ORC A
- An ab initio, DFT and semiempirical SCF-MO package -
Version 4.0
Design and Scientific Directorship:

Frank Neese
Technical Directorship:
Frank Wennmohs
Max-Planck-Institute for Chemical Energy Conversion
Stiftstr. 34-36, 45470 M
ulheim a. d. Ruhr, Germany
[email protected]
With contributions from:

Daniel Aravena, Michael Atanasov, Ute Becker, Dmytro Bykov,
Vijay G. Chilkuri, Dipayan Datta, Achintya Kumar Dutta,
Dmitry Ganyushin, Yang Guo, Andreas Hansen, Lee Huntington,
R
obert Izs
ak, Christian Kollmar, Simone Kossmann, Martin Krupi cka,
Dagmar Lenk, Dimitrios G. Liakos, Dimitrios Manganas,
Dimitrios A. Pantazis, Taras Petrenko, Peter Pinski, Christoph Reimann,
Marius Retegan, Christoph Riplinger, Tobias Risthaus, Michael Roemelt,
Masaaki Saitow, Barbara Sandh ofer, Igor Schapiro,
Kantharuban Sivalingam, Georgi Stoychev, Boris Wezisla
And contributions from our collaborators:

Mih
aly K
allay, Stefan Grimme, Edward Valeev, Garnet Chan, Jiri Pittner,
Martin Brehm
Additional contributions to the manual from:
Giovanni Bistoni, Wolfgang Schneider
Initial conversion from the original Word document to LATEX:

Sarah Lehnhausen
I
Contents
Contents I
List of Figures XI
List of Tables XV
1 Foreword 2017 (Version 4.0) XVI
2 ORCA 4.0 Changes XIX

2.1 New Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIX
2.2 Changes of Default Settings & Parameters . . . . . . . . . . . . . . . . . . . . . . . . XXI
2.3 Important Bugfixes with respect to ORCA 3.0.3 . . . . . . . . . . . . . . . . . . . . XXI
3 FAQ frequently asked questions XXIII
4 General Information 1
4.1 Program Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
4.2 Units and Conversion Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
5 Calling the Program (Serial and Parallel) 3

5.1 Calling the Programm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
5.2 Hints on the Use of Parallel ORCA . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
6 General Structure of the Input File 9

6.1 Input Blocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
6.2 Keyword Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
6.2.1 Main Methods and Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
6.2.2 Density Functional Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
6.3 Basis Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
6.3.1 Standard basis set library . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
6.3.2 Use of scalar relativistic basis sets . . . . . . . . . . . . . . . . . . . . . . . . 30
6.3.3 Effective Core Potentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
6.4 ORCA and Symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
6.5 Jobs with Multiple Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
7 Input of Coordinates 36
7.1 Reading coordinates from the input file . . . . . . . . . . . . . . . . . . . . . . . . . 36
7.2 Reading coordinates from external files . . . . . . . . . . . . . . . . . . . . . . . . . . 38
7.3 Special definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
II Contents
8 Running Typical Calculations 40

8.1 Single Point Energies and Gradients . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
8.1.1 Hartree-Fock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
8.1.1.1 Standard Single Points . . . . . . . . . . . . . . . . . . . . . . . . . 40
8.1.1.2 Basis Set Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
8.1.1.3 SCF and Symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
8.1.2 MP2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
8.1.2.1 MP2 and RI-MP2 Energies . . . . . . . . . . . . . . . . . . . . . . . 45
8.1.2.2 Frozen Core Options . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
8.1.2.3 Orbital Optimized MP2 Methods . . . . . . . . . . . . . . . . . . . 48
8.1.2.4 MP2 and RI-MP2 Gradients and Hessians . . . . . . . . . . . . . . 49
8.1.2.5 MP2 Properties, Densities and Natural Orbitals . . . . . . . . . . . 50
8.1.2.6 Explicitly correlated MP2 calculations . . . . . . . . . . . . . . . . . 52
8.1.2.7 Local MP2 calculations . . . . . . . . . . . . . . . . . . . . . . . . . 52
8.1.3 Coupled-Cluster and Coupled-Pair Methods . . . . . . . . . . . . . . . . . . . 54
8.1.3.1 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
8.1.3.2 Coupled-Cluster Densities . . . . . . . . . . . . . . . . . . . . . . . . 58
8.1.3.3 Static versus Dynamic Correlation . . . . . . . . . . . . . . . . . . . 60
8.1.3.4 Basis Sets for Correlated Calculations. The case of ANOs. . . . . . 63
8.1.3.5 Automatic exptrapolation to the basis set limit . . . . . . . . . . . . 66
8.1.3.6 Explicitly Correlated MP2 and CCSD(T) Calculations . . . . . . . . 71
8.1.3.7 Frozen Core Options . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
8.1.3.8 Local Coupled Pair and Coupled-Cluster Calculations . . . . . . . . 74
8.1.3.9 Cluster in molecules (CIM) . . . . . . . . . . . . . . . . . . . . . . . 85
8.1.3.10 Arbitrary Order Coupled-Cluster Calculations . . . . . . . . . . . . 86
8.1.4 Density Functional Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
8.1.4.1 Standard Density Functional Calculations . . . . . . . . . . . . . . . 87
8.1.4.2 DFT Calculations with RI . . . . . . . . . . . . . . . . . . . . . . . 87
8.1.4.3 HartreeFock and Hybrid DFT Calculations with RIJCOSX . . . . 90
8.1.4.4 HartreeFock and Hybrid DFT Calculations with RI-JK . . . . . . 91
8.1.4.5 DFT Calculations with Second Order Perturbative Correction (Double-
Hybrid Functionals) . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
8.1.4.6 DFT Calculations with Atom-pairwise Dispersion Correction . . . . 92
8.1.4.7 DFT Calculations with Range-Separated Hybrid Functionals . . . . 93
8.1.5 Quadratic Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
8.1.6 Counterpoise Correction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
8.1.7 Complete Active Space Self-Consistent Field Method . . . . . . . . . . . . . . 98
8.1.7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
8.1.7.2 A simple Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
8.1.7.3 Starting Orbitals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
8.1.7.4 CASSCF and Symmetry . . . . . . . . . . . . . . . . . . . . . . . . 108
8.1.7.5 RI, RIJCOSX and RIJK approximations for CASSCF . . . . . . . . 114
8.1.7.6 Breaking Chemical Bonds . . . . . . . . . . . . . . . . . . . . . . . . 115
8.1.7.7 Excited States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
8.1.7.8 CASSCF Natural Orbitals as Input for Coupled-Cluster Calculations123
8.1.8 N-Electron Valence State Perturbation Theory (NEVPT2) . . . . . . . . . . . 128
8.1.9 Full Configuration Interaction Energies and Properties . . . . . . . . . . . . . 131
8.1.10 Scalar Relativistic SCF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
8.1.10.1 Douglas-Kroll-Hess . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Contents III
8.1.10.2 ZORA and IORA . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133

8.1.11 Efficient Calculations with Atomic Natural Orbitals . . . . . . . . . . . . . . 134
8.2 Geometry Optimizations, Surface Scans, Transition States, MECPs . . . . . . . . . . 136
8.2.1 Geometry Optimizations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
8.2.2 Numerical Gradients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
8.2.3 Some Notes and Tricks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
8.2.4 Initial Hessian for Minimization . . . . . . . . . . . . . . . . . . . . . . . . . . 138
8.2.5 Coordinate Systems for Optimizations . . . . . . . . . . . . . . . . . . . . . . 139
8.2.6 Constrained Optimizations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
8.2.7 Constrained Optimizations for Molecular Clusters (Fragment Optimization) . 142
8.2.8 Relaxed Surface Scans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
8.2.9 Multiple XYZ File Scans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
8.2.10 Transition States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
8.2.10.1 Introduction to Transition State Searches . . . . . . . . . . . . . . . 148
8.2.10.2 Hessians for Transition State Calculations . . . . . . . . . . . . . . . 151
8.2.10.3 Special Coordinates for Transition State Optimizations . . . . . . . 152
8.2.11 MECP Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
8.2.12 Constant External Force - Mechanochemistry . . . . . . . . . . . . . . . . . . 155
8.3 Vibrational Frequencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
8.4 Reaction Energies from Compound Jobs . . . . . . . . . . . . . . . . . . . . . . . . . 159
8.5 Excited States with RPA, CIS, CIS(D), ROCIS and TD-DFT . . . . . . . . . . . . . 161
8.5.1 General Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
8.5.2 Use of TD-DFT for the Calculation of X-ray Absorption Spectra . . . . . . . 162
8.5.3 Excited State Geometry Optimization . . . . . . . . . . . . . . . . . . . . . . 166
8.5.4 Doubles Correction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
8.5.5 Restricted Open-shell CIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
8.6 SCF Stability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
8.7 Excited States with EOM-CCSD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
8.7.1 General Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
8.7.2 Capabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
8.8 Excited States with STEOM-CCSD . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
8.8.1 General Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
8.8.2 Capabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
8.9 Excited States with PNO based coupled cluster methods . . . . . . . . . . . . . . . . 173
8.9.1 General Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
8.9.2 Capabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
8.10 Multireference Configuration Interaction and Pertubation Theory . . . . . . . . . . . 175
8.10.1 Introductory Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
8.10.2 A Tutorial Type Example of a MR Calculation . . . . . . . . . . . . . . . . . 179
8.10.3 Excitation Energies between Different Multiplicities . . . . . . . . . . . . . . 195
8.10.4 Correlation Energies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
8.10.5 Thresholds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
8.10.5.1 Reference Values for Total Energies . . . . . . . . . . . . . . . . . . 198
8.10.5.2 Convergence of Single Reference Approaches with Respect to Tsel . 198
8.10.5.3 Convergence of Multireference Approaches with Respect to Tpre . . 199
8.10.6 Energy Differences - Bond Breaking . . . . . . . . . . . . . . . . . . . . . . . 200
8.10.7 Energy Differences - Spin Flipping . . . . . . . . . . . . . . . . . . . . . . . . 203
8.10.8 Potential Energy Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
8.10.9 Multireference Systems - Ozone . . . . . . . . . . . . . . . . . . . . . . . . . . 209
IV Contents
8.10.10 Size Consistency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210

8.10.11 Efficient MR-MP2 Calculations for Larger Molecules . . . . . . . . . . . . . . 211
8.11 MR-EOM-CC: Multireference Equation of Motion Coupled-Cluster . . . . . . . . . . 213
8.11.1 A Simple MR-EOM Calculation . . . . . . . . . . . . . . . . . . . . . . . . . 214
8.11.2 Capabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
8.12 Solvation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
8.13 Calculation of Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
8.13.1 Population Analysis and Related Things . . . . . . . . . . . . . . . . . . . . . 222
8.13.2 Absorption and Fluorescence Bandshapes . . . . . . . . . . . . . . . . . . . . 226
8.13.3 IR/Raman Spectra, Vibrational Modes and Isotope Shifts . . . . . . . . . . . 233
8.13.3.1 IR Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
8.13.3.2 Raman Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
8.13.3.3 Resonance Raman Spectra . . . . . . . . . . . . . . . . . . . . . . . 236
8.13.3.4 NRVS Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
8.13.3.5 Animation of Vibrational Modes . . . . . . . . . . . . . . . . . . . . 238
8.13.3.6 Isotope Shifts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
8.13.4 Thermochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
8.13.5 Anharmonic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
8.13.6 Electrical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
8.13.7 NMR Chemical Shifts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
8.13.8 Hyperfine and Quadrupole Couplings . . . . . . . . . . . . . . . . . . . . . . 254
8.13.9 The EPR g-Tensor and the Zero-Field Splitting Tensor . . . . . . . . . . . . . 257
8.13.10 Mossbauer Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
8.13.11 Broken-Symmetry Wavefunctions and Exchange Couplings . . . . . . . . . . 261
8.13.12 Natural Orbitals for Chemical Valence . . . . . . . . . . . . . . . . . . . . . . 264
8.13.13 Local Energy Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
8.14 QM/MM Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
8.14.1 ORCA and Gromacs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
8.14.2 ORCA and pDynamo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
8.14.3 ORCA and NAMD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
9 Detailed Documentation 277

9.1 More on Coordinate Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
9.1.1 Fragment Specification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
9.1.2 Defining Geometry Parameters and Scanning Potential Energy Surfaces . . . 277
9.1.3 Inclusion of Point Charges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
9.2 Choice of Computational Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
9.2.1 Features Common to All Calculations . . . . . . . . . . . . . . . . . . . . . . 282
9.2.2 Density Functional Calculations . . . . . . . . . . . . . . . . . . . . . . . . . 283
9.2.2.1 Choice of Functional . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
9.2.2.2 Choice of Integration Grid . . . . . . . . . . . . . . . . . . . . . . . 291
9.2.2.3 Using the RI-J Approximation to the Coulomb Part . . . . . . . . . 296
9.2.2.4 The Split-RI-J Coulomb Approximation . . . . . . . . . . . . . . . . 298
9.2.2.5 Using the RI Approximation for Hartree-Fock and Hybrid DFT
(RIJONX) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
9.2.2.6 Using the RI Approximation for Hartree-Fock and Hybrid DFT
(RIJCOSX) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
9.2.2.7 Choice of the COSX Grid and Convergence Issues . . . . . . . . . . 302
9.2.2.8 Improved Analytical Evaluation of the Coulomb Term: Split-J . . . 304
Contents V
9.2.2.9 The Standard Computational Levels . . . . . . . . . . . . . . . . . . 305

9.2.2.10 Treatment of Dispersion Interactions with DFT-D3 . . . . . . . . . 305
9.2.2.11 DFT Calculations with the Non-Local, Density Dependent Disper-
sion Correction: DFT-NL . . . . . . . . . . . . . . . . . . . . . . . . 310
9.2.2.12 DFT and HF Calculations with the geometrical Counterpoise Cor-
rection: gCP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
9.2.2.13 HF-3c: Hartree-Fock with three corrections . . . . . . . . . . . . . . 318
9.2.2.14 PBEh-3c: A PBE hybrid density functional with small AO basis
set and two corrections . . . . . . . . . . . . . . . . . . . . . . . . . 320
9.2.3 Semiempirical Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
9.3 Choice of Basis Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
9.3.1 Built-in Basis Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
9.3.2 Assigning or Adding Basis Functions to an Element . . . . . . . . . . . . . . 331
9.3.3 Assigning or Adding Basis Functions to Individual Atoms . . . . . . . . . . . 333
9.3.4 Linear Dependence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
9.3.5 Reading Orbital and Auxiliary Basis Sets from a File . . . . . . . . . . . . . 335
9.3.6 Advanced Specification of Effective Core Potentials . . . . . . . . . . . . . . . 336
9.3.7 Embedding Potentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
9.4 Choice of Initial Guess and Restart of SCF Calculations . . . . . . . . . . . . . . . . 339
9.4.1 AutoStart feature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
9.4.2 One Electron Matrix Guess . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
9.4.3 Basis Set Projection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
9.4.4 PModel Guess . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
9.4.5 H uckel and PAtom Guesses . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
9.4.6 Restarting SCF Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . 341
9.4.7 Changing the Order of Initial Guess MOs and Breaking the Initial Guess
Symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
9.5 SCF Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343
9.5.1 Convergence Tolerances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343
9.5.2 Dynamic and Static Damping . . . . . . . . . . . . . . . . . . . . . . . . . . . 344
9.5.3 Level Shifting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
9.5.4 Direct Inversion in Iterative Subspace (DIIS) . . . . . . . . . . . . . . . . . . 346
9.5.5 An alternative DIIS algorithm: KDIIS . . . . . . . . . . . . . . . . . . . . . . 346
9.5.6 Approximate Second Order SCF (SOSCF) . . . . . . . . . . . . . . . . . . . . 347
9.5.7 Full Second Order SCF (Newton-Raphson SCF) . . . . . . . . . . . . . . . . 347
9.5.8 Finite Temperature HF/KS-DFT . . . . . . . . . . . . . . . . . . . . . . . . . 349
9.5.8.1 Fractional Occupation Numbers . . . . . . . . . . . . . . . . . . . . 350
9.5.8.2 Fractional Occupation Number Weighted Electron Density (FOD) . 350
9.6 Choice of Wavefunction and Integral Handling . . . . . . . . . . . . . . . . . . . . . 353
9.6.1 Choice of Wavefunction Type . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
9.6.2 UHF Natural Orbitals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
9.6.3 Integral Handling (Conventional, Direct, SemiDirect) . . . . . . . . . . . . . . 355
9.7 SCF Stability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
9.8 Frozen Core Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
9.9 The Second Order Many Body Pertubation Theory Module (MP2) . . . . . . . . . . 362
9.9.1 Standard MP2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
9.9.2 RI-MP2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
9.9.3 Double-Hybrid Density Functional Theory . . . . . . . . . . . . . . . . . . 365
9.9.4 Orbital Optimized MP2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
VI Contents
9.9.5 RIJCOSX-RI-MP2 Gradients . . . . . . . . . . . . . . . . . . . . . . . . . . . 374

9.9.6 MP2 and RI-MP2 Second Derivatives . . . . . . . . . . . . . . . . . . . . . . 374
9.9.7 Local MP2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
9.10 The Single Reference Correlation Module . . . . . . . . . . . . . . . . . . . . . . . . 378
9.10.1 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378
9.10.2 Closed-Shell Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
9.10.3 Open-Shell Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386
9.10.4 Local correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386
9.10.4.1 Including core orbitals in the correlation treatment . . . . . . . . . 392
9.10.5 Hilbert space multireference coupled-cluster approaches . . . . . . . . . . . . 394
9.10.6 The singles Fock term . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396
9.10.7 Use of the MDCI Module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
9.11 The Complete Active Space Self-Consistent Field (CASSCF) Module . . . . . . . . . 400
9.11.1 General Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 400
9.11.2 CASSCF Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 420
9.11.3 Fully Variational Spin-Orbit Coupled CASSCF . . . . . . . . . . . . . . . . . 421
9.12 N-Electron Valence State Pertubation Theory . . . . . . . . . . . . . . . . . . . . . . 427
9.13 Density Matrix Renormalization Group . . . . . . . . . . . . . . . . . . . . . . . . . 438
9.13.1 Technical capabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 439
9.13.2 How to cite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 439
9.13.3 Overview of BLOCK input and calculations . . . . . . . . . . . . . . . . . . . 439
9.13.4 Standard commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440
9.13.4.1 Orbital optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . 441
9.13.4.2 Advanced options . . . . . . . . . . . . . . . . . . . . . . . . . . . . 441
9.13.4.3 Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
9.13.4.4 Complete set of BLOCK options . . . . . . . . . . . . . . . . . . . . . 444
9.13.5 Appendix: Porphine -active space calculation . . . . . . . . . . . . . . . . . 445
9.14 Relativistic Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
9.14.1 Approximate Relativistic Hamiltonians . . . . . . . . . . . . . . . . . . . . . 450
9.14.2 The Regular Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451
9.14.3 The Douglas-Kroll-Hess Method . . . . . . . . . . . . . . . . . . . . . . . . . 453
9.14.4 Picture-Change Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454
9.14.5 Finite Nucleus Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456
9.14.6 Basis Sets in Relativistic Calculations . . . . . . . . . . . . . . . . . . . . . . 456
9.15 Approximate Full CI Calculations in Subspace: ICE-CI . . . . . . . . . . . . . . . . 457
9.15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457
9.15.2 The ICE-CI and CIPSI Algorithms . . . . . . . . . . . . . . . . . . . . . . . . 458
9.15.3 A Simple Example Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . 459
9.15.4 Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
9.15.5 Scaling behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
9.15.6 Accuracy of the Wavefunction . . . . . . . . . . . . . . . . . . . . . . . . . . . 463
9.15.7 Potential Energy Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
9.15.8 Excited States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466
9.15.9 Tips and Tricks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
9.15.10 Large-scale approximate CASSCF: ICE-SCF . . . . . . . . . . . . . . . . . . 471
9.15.11 The entire input block explained . . . . . . . . . . . . . . . . . . . . . . . . . 472
9.15.12 A Technical Note: orca cclib . . . . . . . . . . . . . . . . . . . . . . . . . . 473
9.16 CI methods using generated code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
9.16.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
Contents VII
9.16.2 Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474

9.16.3 Fully internally contracted MRCI . . . . . . . . . . . . . . . . . . . . . . . . . 476
9.17 Geometry Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
9.17.1 Input Options and General Considerations . . . . . . . . . . . . . . . . . . . 477
9.17.2 Transition State Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
9.17.3 Minimum Energy Crossing Points . . . . . . . . . . . . . . . . . . . . . . . . 491
9.17.4 Numerical Gradients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
9.17.5 ORCA as External Optimizer . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
9.18 Excited States via RPA, CIS and TD-DFT . . . . . . . . . . . . . . . . . . . . . . . 495
9.18.1 General Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
9.18.2 Semiempirical Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
9.18.3 Hartree-Fock Wavefunctions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
9.18.4 Non-Hybrid and Hybrid DFT . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
9.18.5 Simplified TDA and TD-DFT . . . . . . . . . . . . . . . . . . . . . . . . . . . 502
9.18.5.1 Theoretical Background . . . . . . . . . . . . . . . . . . . . . . . . . 502
9.18.5.2 Calculation Set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503
9.18.6 Double-hybrid functionals and Doubles Correction . . . . . . . . . . . . . . . 506
9.18.7 Natural Transition Orbitals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 508
9.18.8 Computational Aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 510
9.18.8.1 RI Approximation (AO-Basis) . . . . . . . . . . . . . . . . . . . . . 510
9.18.8.2 RI Approximation (MO-Basis) . . . . . . . . . . . . . . . . . . . . . 510
9.18.8.3 Integral Handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 511
9.18.8.4 Valence versus Rydberg States . . . . . . . . . . . . . . . . . . . . . 512
9.18.8.5 Restrictions for Range-Separated Density Functionals . . . . . . . . 512
9.18.8.6 Asymptotatically Corrected Density Functionals . . . . . . . . . . . 513
9.18.8.7 Potential Energy Surface Scans . . . . . . . . . . . . . . . . . . . . . 514
9.18.8.8 Potential Energy Surface Scans along Normal Coordinates . . . . . 515
9.18.8.9 Normal Mode Scan Calculations Between Different Structures . . . 520
9.19 Excited States via ROCIS and DFT/ROCIS . . . . . . . . . . . . . . . . . . . . . . . 521
9.19.1 General Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
9.19.2 Transition Metal L-Edges with ROCIS or DFT/ROCIS . . . . . . . . . . . . 528
9.19.3 Natural Transition Orbitals/ Natural Difference Orbitals . . . . . . . . . . . . 534
9.19.4 Resonant Inelastic Scattering Spectroscopy . . . . . . . . . . . . . . . . . . . 535
9.19.4.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 535
9.19.4.2 Processing the spectra with orca mapspc . . . . . . . . . . . . . . . 539
9.19.4.3 Generating Cuts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 542
9.19.5 Keyword List . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 543
9.20 Excited States via EOM-CCSD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
9.20.2 Memory Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 547
9.20.3 Initial Guess . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
9.20.4 Hamiltonian Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549
9.20.5 Solution of the (Nonsymmetric) Eigenproblem . . . . . . . . . . . . . . . . . 549
9.20.6 Convergence, Restart, Preconditioning and Subspace Expansion . . . . . . . 550
9.20.7 Properties in the RHF EOM implementation . . . . . . . . . . . . . . . . . . 551
9.20.8 Some tips and tricks for EOM-CC calculation . . . . . . . . . . . . . . . . . . 552
9.21 Excited States via STEOM-CCSD . . . . . . . . . . . . . . . . . . . . . . . . . . . . 552
9.21.2 Selection of Active space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 554
VIII Contents
9.21.3 The reliabilty of the calculated excitation energy . . . . . . . . . . . . . . . . 555

9.21.4 Removal of IP and EA states with double excitation character . . . . . . . . 557
9.21.5 Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 558
9.21.6 Some tips and tricks for STEOMCC calculations . . . . . . . . . . . . . . . . 558
9.22 Excited States using PNO based coupled cluster . . . . . . . . . . . . . . . . . . . . 559
9.22.2 Reference State Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559
9.22.3 Use of Local Orbitals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 560
9.22.4 Some tips and tricks for bt-PNO calculations . . . . . . . . . . . . . . . . . . 560
9.23 The Multireference Correlation Module . . . . . . . . . . . . . . . . . . . . . . . . . 561
9.23.2 Properties Calculation Using the SOC Submodule . . . . . . . . . . . . . . . 567
9.23.2.1 Zero-Field Splitting . . . . . . . . . . . . . . . . . . . . . . . . . . . 567
9.23.2.2 Local Zero-Field Splitting . . . . . . . . . . . . . . . . . . . . . . . . 575
9.23.2.3 g-Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 578
9.23.2.4 Magnetization and Magnetic Susceptibility . . . . . . . . . . . . . . 583
9.23.2.5 MCD and Absorption Spectra . . . . . . . . . . . . . . . . . . . . . 585
9.23.2.6 Addition of Magnetic Fields . . . . . . . . . . . . . . . . . . . . . . 589
9.23.2.7 Relativistic Picture Change in Douglas-Kroll-Hess SOC and Zeeman
Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 591
9.24 Multireference Equation of Motion Coupled-Cluster (MR-EOM-CC) Theory . . . . . 592
9.24.1 The Steps Required to Run an MR-EOM Calculation . . . . . . . . . . . . . 595
9.24.1.1 State-Averaged CASSCF Calculation . . . . . . . . . . . . . . . . . 595
9.24.1.2 Selection of the States to Include in the MR-EOM Calculation . . . 596
9.24.1.3 Running the MR-EOM Calculation . . . . . . . . . . . . . . . . . . 598
9.24.2 Approximate Inclusion of Spin-Orbit Coupling Effects in MR-EOM Calcula-
tions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 600
9.24.3 A Projection/Singular PT Scheme to Overcome Convergence Issues in the T
Amplitude Iterations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604
9.24.4 An Orbital Selection Scheme for More Efficient Calculations of Excitation
Spectra with MR-EOM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 608
9.24.5 Nearly Size Consistent Results with MR-EOM by Employing an MR-CEPA(0)
Shift in the Final Diagonalization Procedure . . . . . . . . . . . . . . . . . . 615
9.25 Simulation and Fit of Vibronic Structure in Electronic Spectra, Resonance Raman
Excitation Profiles and Spectra with the orca asa Program . . . . . . . . . . . . . . 618
9.25.1 General Description of the Program . . . . . . . . . . . . . . . . . . . . . . . 619
9.25.2 Spectral Simulation Procedures: Input Structure and Model Parameters . . . 619
9.25.2.1 Example: Simple Mode . . . . . . . . . . . . . . . . . . . . . . . . . 619
9.25.2.2 Example: Modelling of Absorption and Fluorescence Spectra within
the IMDHO Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 621
9.25.2.3 Example: Modelling of Absorption and Fluorescence Spectra within
the IMDHOFA Model . . . . . . . . . . . . . . . . . . . . . . . . . . 623
9.25.2.4 Example: Modelling of Effective Broadening, Effective Stokes Shift
and Temperature Effects in Absorption and Fluorescence Spectra
within the IMDHO Model . . . . . . . . . . . . . . . . . . . . . . . 625
9.25.2.5 Example: Modelling of Absorption and Resonance Raman Spectra
for the 1-1 Ag 1-1 Bu Transition in trans-1,3,5-Hexatriene . . . . . 628
9.25.2.6 Example: Modelling of Absorption Spectrum and Resonance Ra-
man Profiles for the 1-1 Ag 1-1 Bu Transition in trans-1,3,5-Hexatriene632
Contents IX
9.25.3 Fitting of Experimental Spectra . . . . . . . . . . . . . . . . . . . . . . . . . 634

9.25.3.1 Example: Gauss-Fit of Absorption Spectrum . . . . . . . . . . . . . 634
9.25.3.2 Example: Fit of Absorption and Resonance Raman Spectra for 1-
1 A 1-1 B Transition in trans-1,3,5-Hexatriene . . . . . . . . . . 641
g u
9.25.3.3 Example: Single-Mode Fit of Absorption and Fluorescence Spectra
for 1-1 Ag 1-1 B2u Transition in Tetracene . . . . . . . . . . . . . . 648
9.25.4 Quantum-Chemically Assisted Simulations and Fits of Optical Bandshapes
and Resonance Raman Intensities . . . . . . . . . . . . . . . . . . . . . . . . . 652
9.25.4.1 Example: Quantum-Chemically Assisted Analysis and Fit of the
Absorption and Resonance Raman Spectra for 1-1 Ag 1-1 Bu Tran-
sition in trans-1,3,5-Hexatriene . . . . . . . . . . . . . . . . . . . . . 652
9.25.4.2 Important Notes about Proper Comparison of Experimental and
Quantum Chemically Calculated Resonance Raman Spectra . . . . 661
9.25.4.3 Example: Normal Mode Scan Calculations of Model Parameters for
1-1 Ag 1-1 Bu Transition in trans-1,3,5-Hexatriene . . . . . . . . . 661
9.26 Ab initio Molecular Dynamics Simulations . . . . . . . . . . . . . . . . . . . . . . . . 665
9.26.1 Input Format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 665
9.26.2 Command List . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 669
9.26.3 Scientific Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 671
9.26.3.1 Time Integration and Equations of Motion . . . . . . . . . . . . . . 672
9.26.3.2 Velocity Initialization . . . . . . . . . . . . . . . . . . . . . . . . . . 673
9.26.3.3 Thermostats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 674
9.27 Implicit Solvation Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 675
9.27.1 The Conductor-like Polarizable Continuum Model (C-PCM) . . . . . . . . . . 675
9.27.2 The Conductor-like Screening Solvation Model (COSMO) . . . . . . . . . . . 677
9.27.3 The SMD Solvation Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 677
9.28 Calculation of Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 681
9.28.1 Electric Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 681
9.28.2 The Spin-Orbit Coupling Operator . . . . . . . . . . . . . . . . . . . . . . . . 682
9.28.2.1 Exclusion of Atomic Centers . . . . . . . . . . . . . . . . . . . . . . 685
9.28.3 The EPR/NMR Module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 686
9.28.3.1 Hyperfine and Quadrupole Couplings . . . . . . . . . . . . . . . . . 688
9.28.3.2 The g-Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 690
9.28.3.3 Zero-Field-Splitting . . . . . . . . . . . . . . . . . . . . . . . . . . . 691
9.28.3.4 General Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 692
9.28.4 Local Energy Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . 698
9.29 Natural Bond Orbital (NBO) Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 699
9.29.1 NBO Deletions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 708
9.29.2 NBO for Post-HF Densities . . . . . . . . . . . . . . . . . . . . . . . . . . . . 711
9.30 Population Analyses and Control of Output . . . . . . . . . . . . . . . . . . . . . . . 712
9.30.1 Controlling Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 712
9.30.2 Mulliken Population Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 716
9.30.3 Lowdin Population Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 718
9.30.4 Mayer Population Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 719
9.30.5 Natural Population Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 720
9.30.6 Density of the Effectively Unpaired Electrons . . . . . . . . . . . . . . . . . . 720
9.30.7 Local Spin Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 723
9.30.8 UNO Orbital Printing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 729
X Contents
9.31 Orbital and Density Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 730

9.31.1 Contour Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 730
9.31.2 Surface Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 732
9.31.2.1 General Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 732
9.31.2.2 FOD plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 734
9.31.2.3 Interface to gOpenMol . . . . . . . . . . . . . . . . . . . . . . . . . 736
9.31.2.4 Interface to Molekel . . . . . . . . . . . . . . . . . . . . . . . . . . . 736
9.32 Utility Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 737
9.32.1 orca mapspc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 737
9.32.2 orca chelpg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 740
9.32.3 orca pltvib . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 741
9.32.4 orca vib . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 741
9.32.5 orca loc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 741
9.32.6 orca blockf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 743
9.32.7 orca plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 743
9.32.8 orca 2mkl: Old Molekel as well as Molden inputs . . . . . . . . . . . . . . . . 744
9.32.9 orca 2aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 744
9.32.10 orca vpot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 745
9.32.11 orca euler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 745
9.32.12 orca exportbasis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 747
9.32.13 orca eca . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 748
10 Some Tips and Tricks 749

10.1 Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 749
10.2 Cost versus Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 749
10.3 Converging SCF Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 752
10.4 Choice of Theoretical Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 754
11 Publications Related to ORCA 756
Bibliography 783
XI
List of Figures
8.1 A rigid scan along the twisting coordinate of C2 H4 . The inset shows the T1 diagnostic
for the CCSD calculation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
8.2 Comparison of the CCSD(T) and MRACPF total energies of the C2 H4 along the
twisting coordinate. The inset shows the difference E(MRACPF)-E(CCSD(T)). . . . 61
8.3 Potential energy surface of the F2 molecule calculated with some single-reference
methods and compared to the MRACPF reference. . . . . . . . . . . . . . . . . . . . 63
8.4 Error in mEh for various basis sets for highly correlated calculations relative to the
ano-pVQZ basis set. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
8.5 Scaling behaviour of RHF, canonical and DLPNO-CCSD(T) . . . . . . . . . . . . . . 76
8.6 Structure of the Crambin protein - the first protein to be treated with a CCSD(T)
level ab initio method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
8.7 Computational times of RIJCOSX-UHF and UHF-DLPNO-CCSD for the linear
alkane chains (Cn Hn+2 ) in triplet state with def2-TZVPP basis and default frozen
core settings. 4 CPU cores and 128 GB of memory were used on a single cluster node. 79
8.8 Ni-Fe active center in the [NiFe] Hydrogenase in its second-coordination sphere. The
whole model system is composed of 180 atoms. . . . . . . . . . . . . . . . . . . . . . 80
8.9 A model compound for the OEC in the S2 state of photosystem II which is comprised
of 238 atoms. In its high-spin state, the OEC possesses 13 SOMOs in total. . . . . . 81
8.10 Potential Energy Surface of the H2 molecule from RHF, UHF and CASSCF(2,2)
calculations (SVP basis). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
8.11 Potential Energy Surface of the N2 molecule from CASSCF(6,6) calculations (SVP
basis). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
8.12 State averaged CASSCF(2,2) calculations on H2 (two singlets, one triplet; SVP basis).119
8.13 State averaged CASSCF(2,2) calculations on C2 H4 (two singlets, one triplet; SV(P)
basis). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
8.14 Potential Energy Surface of the N2 molecule from CASSCF(6,6) and NEVPT2 cal-
culations (def2-SVP). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
8.15 Relaxed surface scan for the H-atom abstraction from CH4 by OH-radical (B3LYP/SV(P)).
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
8.16 Parameter scan for the quartet and sextet state of [FeO]+ (B3LYP/SV(P)). . . . . . 154
8.17 Potential energy surfaces for some low-lying states of CH using the MRCI+Q method208
8.18 Frontier MOs of the Ozone Molecule. . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
8.19 2D potential energy surface for the O3 molecule calculated with different methods. . 211
8.20 Rotation of stilbene around the central double bond using a CASSCF(2,2) reference
and correlating the reference with MR-MP2. . . . . . . . . . . . . . . . . . . . . . . . 212
8.21 The and orbitals of the CO molecule obtained from the interface of ORCA to
gOpenMol. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
8.22 The and -MOs of CO as visualized by Molekel. . . . . . . . . . . . . . . . . . . . 225
8.23 The predicted IR spectrum of the H2 CO molecule using the numerical frequency
routine of ORCA and the tool orca mapspc to create the spectrum. . . . . . . . . . 234
XII List of Figures
8.24 Calculated Raman spectrum for H2 CO at the STO-3G level using the numerical
frequency routine of ORCA and the tool orca mapspc to create the spectrum. . . . . 236
8.25 Experimental, fitted and simulated NRVS spectrum of the Fe(III)-azide complex
obtained at the BP86/TZVP level (T = 20 K). . . . . . . . . . . . . . . . . . . . . . 237
8.26 Theoretical IR spectrum with the shapes of vibrations dominating the IR intensity
and NRVS scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
8.27 Nuclear vibrations for H2 CO with the shape of each vibration and its frequency
indicated . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
8.28 The 1395 cm1 mode of the H2 CO molecule as obtained from the interface of ORCA
to gOpenMol and the orca pltvib tool to create the animation file. . . . . . . . . . . 241
8.29 An idealized Mossbauer spectrum showing both the isomer shift, , and the quadrupole
splitting, EQ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
9.1 Graphical description of the Range-Separation ansatz. . . . . . . . . . . . . . . . . . 290

9.2 Scaling of the DLPNO-MP2 method with default thresholds for linear alkane chains
in def2-TZVP basis. Shown are also the times for the corresponding Hartree-Fock
calculations with RIJCOSX and for RI-MP2. . . . . . . . . . . . . . . . . . . . . . . 375
9.3 Orbitals of the active space for the CASSCF(6,6) calculation of H2 CO. . . . . . . . . 408
9.4 Structure of the FeC72 N2 H100 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 431
9.5 NEVPT2 and QDNEVPT2 Li-F dissociation curves of the ground and first excited
states for a CAS(2,2) reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436
9.6 Flowchart of the ICE-CI procedure. . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
9.7 An ICE-CI calculation on the water molecule in the cc-pVDZ basis (1s frozen) . . . 461
9.8 Convergence of the ICE-CI procedure towards the full CI results for a test set of 21
full CI energy. Shown is the RMS error relative to the Full CI results. The corre-
sponding errors for various coupled-cluster variants is shown by broken horizontal
lines. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
9.9 Polyene chains used for scaling calculations. . . . . . . . . . . . . . . . . . . . . . . . 462
9.10 Scaling behavior of ICE-CI for linear polyene chains (Full -electron active space)
as a functions of system size for different generator thresholds. . . . . . . . . . . . . 463
9.11 Convergence of the ICE-CI natural orbital occupation numbers. The upper panel
is showing the Full CI occupation numbers, the lower panel the deviation of the
ICE-CI values from these exact values. For comparison, the CCSD natural orbital
occupation numbers are also provided. . . . . . . . . . . . . . . . . . . . . . . . . . 464
9.12 Potential energy surface of the N2 molecule in the SV basis. For comparison higher
level coupled-cluster results are also shown. . . . . . . . . . . . . . . . . . . . . . . . 465
9.13 Non-parallelity error of ICE-CI for the H2 O molecule in the SV basis. Shown is the
deviation from the full CI value as a function of O-H distance (both bonds stretched).
For comparison, the CCSD(T) curve is also shown . . . . . . . . . . . . . . . . . . . 466
9.14 Analysis of the ICE-CI wavefunction along the O-H dissociation pathway. . . . . . . 467
9.15 Comparison of MP2 natural orbitals and improved virtual orbitals for the ICE-CI
procedure (H2O molecule, cc-pVDZ basis, equilibrium geometry) . . . . . . . . . . . 468
9.16 Automatic active space selection along the H2 O dissociation surface. The reference
curve (blue triangles) is the ICE-CI method for the full orbital space with the default
parameters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
9.17 Deviations of ICE-SCF from CASSCF energies for a selection of molecules (standard
truncation parameters Tgen = 104 and Tvar = 1011 ) . . . . . . . . . . . . . . . . . . 472
9.18 CASSCF and ICE-SCF optimized geometries for methylene and ozone (cc-pVDZ
basis set, default parameters). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
List of Figures XIII
9.19 Structure of the iron-porphyrin used for the prediction of its absorption spectrum. . 500
9.20 The ZINDO/S predicted absorption spectrum of the model iron-porphyrin. The
spectrum has been plotted using the orca mapspc tool. . . . . . . . . . . . . . . . . . 500
9.21 Natural transition orbitals for the pyridine molecule in the S1 and S3 states. . . . . 510
9.22 Result of a potential energy surface scan for the excited states of the CO molecule
using the orca cis module. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515
9.23 Result of a potential energy surface scan along C-C stretching normal coordinate
(mode 13 in the present example) for the excited states of the ethene molecule using
the orca cis module. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
9.24 Comparison of the experimentally observed and calculated ROCIS Fe L-edge of
[FeCl4 ]2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532
9.25 Comparison of the experimentally observed and calculated Ti L-edge of [Cp2 TiCl2 ]. 533
9.26 DFT/ROCIS calculated L3 XAS spectrum of [Fe(Cl)4 ]2 together with NDO analysis.536
9.27 DFT/ROCIS calculated RIXS planes for [Cu(N H3 )4 ]2 . . . . . . . . . . . . . . . . . 541
9.28 Calculated MCD and absorption spectra of [Fe(CN)6 ]3 compared to experimental
spectra. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 590
9.29 Absorption spectrum generated after orca asa run on file example001.inp. . . . . . . 621
9.30 Absorption and fluorescence spectra generated after orca asa run on the file exam-
ple002.inp. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 624
9.31 Absorption and fluorescence spectra generated after orca asa run on the file exam-
ple003.inp. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 625
9.32 Absorption and fluorescence spectra for T=0 K and T=300 K generated after orca asa
run on the file example004.inp. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 627
9.33 Absorption spectrum for T = 300 K generated after orca asa run on the file exam-
ple004.inp. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 628
9.34 Absorption spectrum corresponding to 1 1 Ag 1 1 Bu transition in trans-1,3,5-
hexatriene generated after orca asa run on the file example005.inp. . . . . . . . . . . 630
9.35 Resonance Raman spectra for 3 different excitation energies which fall in resonance
with 1 1 Ag 1 1 Bu transition in trans-1,3,5-hexatriene. . . . . . . . . . . . . . . 630
9.36 Absorption spectrum and resonance Raman profiles of fundamental bands corre-
sponding to 1 1 Ag 1 1 Bu transition in trans-1,3,5-hexatriene. . . . . . . . . . 634
9.37 Experimental absorption spectrum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635
9.38 Comparison of the experimental and fitted absorption spectra corresponding to the
fit run of orca asa on the file example007.inp. . . . . . . . . . . . . . . . . . . . . . . 638
fit run of orca asa on the file example007.001.inp. . . . . . . . . . . . . . . . . . . . . 639
fit run of orca asa on the file example007.002.inp in which equal broadening was
assumed for all electronic bands. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 640
9.41 Experimental Resonance Raman spectrum corresponding to 1-1 Ag 1-1 Bu transi-
tion in trans-1,3,5-hexatriene. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 641
9.42 Experimental and fitted absorption spectrum corresponding to 1-1 Ag 1-1 Bu tran-
sition in trans-1,3,5-hexatriene. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 647
9.43 Deconvoluted absorption and fluorescence spectra of tetracene in cyclohexane upon
the assumption of a single vibronically active mode. . . . . . . . . . . . . . . . . . . 648
9.44 Experimental and calculated at the BHLYP/SV(P) and B3LYP/SV(P) levels of
theory absorption and rR spectra corresponding to 1-1 Ag 1-1 Bu transition in
trans-1,3,5-hexatriene. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 657
9.45 Experimental and theoretical absorption spectra for a single-mode model system. . . 662
XIV List of Figures
9.46 Six NBOs of the H2 CO molecule. Shown are the occupied bonding and orbitals
for C and O, the two oxygen lone-pairs and the two and antibonding orbitals. . 707
9.47 Contour plot of the lowest unoccupied spin down orbital of the H2 CO+ cation radical
in the x, y plane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 731
9.48 The total electron density and the spin density of the H2 CO+ cation radical as
calculated by the RI-BP/VDZP method. . . . . . . . . . . . . . . . . . . . . . . . . . 734
9.49 The orbital of H2 CO as calculated by the RI-BP/VDZP method. . . . . . . . . . 734
9.50 FOD plot at = 0.005 e/Bohr3 (TPSS/def2-TZVP (T = 5000 K) level) for the 1 Ag
ground state of p-benzyne (FOD depicted in yellow). . . . . . . . . . . . . . . . . . . 736
XV
List of Tables
6.1 Main keywords that can be used in the simple input of ORCA. . . . . . . . . . . . . 12
6.2 Density functionals available in ORCA. . . . . . . . . . . . . . . . . . . . . . . . . . 22
6.3 Basis sets available on ORCA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
6.4 Overview of auxiliary basis sets available in ORCA. . . . . . . . . . . . . . . . . . . 29
6.5 Overview of library keywords for ECPs and associated basis sets available in ORCA. 31
8.1 Computer times for solving the coupled-cluster/coupled-pair equations for Serine . . 56
8.2 Computed spectroscopic constants of N2 with coupled-cluster methods. . . . . . . . 57
8.3 Comparison of various basis sets for highly correlated calculations . . . . . . . . . . 64
8.4 Comparison of various basis sets for correlated calculations. . . . . . . . . . . . . . . 65
8.6 Accuracy settings for DLPNO coupled cluster (current version). . . . . . . . . . . . . 83
8.7 Accuracy settings for DLPNO coupled cluster (deprecated 2013 version). . . . . . . . 83
8.11 Comparison of the performance of the MRCI and MDCI modules for a single refer-
ence calculation with the bn-ANO-DZP basis set on the zwitter-ionic form of serine. 177
9.4 Overview of parametrized basis sets. . . . . . . . . . . . . . . . . . . . . . . . . . . . 316

9.5 Composition of the MINIX basis set. . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
9.6 Composition of the def2-mSVP basis set. . . . . . . . . . . . . . . . . . . . . . . . . 321
9.7 Basis sets availability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
9.8 Default values for number of frozen core electrons. . . . . . . . . . . . . . . . . . . . 361
9.9 Accuracy settings for DLPNO-MP2. . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
9.10 Summary of the diagonal shifts used in various singles- and doubles methods. . . . . 384
9.11 Number of chemical core electrons in DLPNO calculation . . . . . . . . . . . . . . . 393
9.12 Comparison of the properties of the fpFW and fFW DKH Hamiltonians. . . . . . . 455
9.13 Keyword list for sTDA and sTD-DFT. . . . . . . . . . . . . . . . . . . . . . . . . . . 503
9.14 Integral handling in various implementations of the (D) correction. . . . . . . . . . . 507
9.15 The details of the various MR-EOM transformations that are considered in the
ORCA implementation of MR-EOM. . . . . . . . . . . . . . . . . . . . . . . . . . . . 593
9.16 Details of the three MR-EOM approaches implemented in ORCA . . . . . . . . . . . 593
9.17 A test for size consistency in MR-EOM. . . . . . . . . . . . . . . . . . . . . . . . . . 617
9.18 A test for size consistency in MR-EOM, using the MR-CEPA(0) shift. . . . . . . . . 618
XVI
1 Foreword 2017 (Version 4.0)
After more than three years of intense development, the ORCA developers are proud to present to you the
fourth version of the program, ORCA 4.0. The present version is a major step forward from previous versions
of the program. We very much hope that it will be a powerful companion to your research endeavors and
that you enjoy using the program as much as we enjoy creating it.
ORCA has now a history of almost twenty years since its shy beginnings as a semi-empirical program package
in the mid/late 1990s. It is fair to say that ORCA has come of age and is now one of the most used quantum
chemistry packages in existence. More than 10000 users are registered in the ORCA forum and actively
participate in discussing the software, its features and how to solve computational chemistry problems using
the powerful methods implemented in ORCA. With the many users at supercomputer centers it is prudent to
speculate that the total number of users is significantly higher than that. A conservative estimate is that
ORCA is definitely within the top five most used quantum chemistry programs and most certainly the one
with the steepest growth rate to date.
Concomitant with the increasing popularity of the program there have been growing demands from industry
to gain access to ORCA. For this reason, we have decided to initiate a company that will distribute ORCA
to industry. These activities will be described fully elsewhere. Here we will only mention that ORCA will
remain free for academic users and we have no plans to ever change this. In fact, the current plans will help to
secure a long-term great future for the project and we are enthusiastically embracing our work on ORCA.
There has been a lot of work to make ORCA 4.0 happen as witnessed by several thousand checkins into our
source code repository that have taken place since the last release. In addition to the new major features,
there has been a lot of work under the hood to ensure that the program parts work smoothly together
and to enhance efficiency and workflow. Examples for this work is the largely improved efficiency of the four
center integral code and the implementation of the full TD-DFT method (including gradients) for hybrid
functionals as well. Furthermore, we have undertaken a dedicated effort to improve the manual and provide
other important practical information, for example, a relatively detailed tutorial that shows how to use the
CASSCF module and the various multireference dynamic correlation methods.
Our method development efforts have ever more strongly concentrated on wavefunction based ab initio
methods. Foremost, the domain based local pair natural orbital (DLPNO) methodology has been largely
improved and extended in ORCA 4.0. In a key paper, we have outlined the concept of Sparse Maps that
provide the infrastructure for the further development of local correlation theories. Using this infrastructure,
we have been able to implement a fully linear scaling version of the DLPNO-CCSD(T) method for closed
shells that is replacing the already highly successful 2013 version. In the process, we have been able to iron out
some remaining glitches, e.g. the domain construction that is now relying on the key concept of differential
overlap integrals and the treatment of core correlation effects in the DLPNO approximation. ORCA 4.0
also features a completely reformulated open-shell DLPNO-CCSD(T) method that remedies some of the
shortcomings of the earlier open-shell LPNO-CCSD method. Furthermore, we have been able to implement
XVII
the DLPNO-CCSD lambda equations for closed shells, thus providing access to coupled cluster level molecular
properties. Very noticeably, the DLPNO methodology has been extended to the multireference case and the
DLPNO-NEVPT2 method is as efficient and accurate as its DLPNO-MP2 counterpart. Work on coupled
cluster excited states in the framework of the equation of motion (EOM) and similarity transformed EOM
(STEOM) has continued and first steps towards their DLPNO implementation have been taken.
Major efforts have been spent to streamline the development of complicated correlation theories using
automated code generation. This will enable us to implement a wide variety of ab initio theories very
efficiently and consistently. The first fruits of this important development are present in ORCA 4.0 via a new
module that is called autoci. This module presently can perform a variety of standard single reference
correlation calculations. However, importantly, it also features a new fully internally contracted MRCI module
that, in many cases, outperforms the existing MRCI module by a large margin. Much more functionality
along these lines is expected to be part of future releases.
Significant progress has also occurred in the ROCIS module, which enables the efficient calculation of many
spectroscopic properties, perhaps most importantly advanced X-ray spectra such as L-edge absorption spectra
and RIXS spectra.
While we always strive for calculating more accurate numbers more quickly, we also aim at the chemical
interpretation of ab initio results by making use of the fact that wavefunctions are intelligible. In ORCA
4.0 two significant new features along these lines are included. These concern the ab initio ligand theory
(AILFT). The AILFT analysis allows the user to extract ligand field parameters in a unique fashion from
CASSCF/NEVPT2 calculations. It has been of major use in a variety of studies, perhaps most prominently in
our work on molecular magnetism. Secondly, the DLPNO-CCSD(T) energy can be decomposed into chemically
meaningful parts such as the dispersion energy. This new method, termed local energy decomposition (LED),
is of major utility, for example, in studying weak intermolecular interactions.
Progress in DFT methods largely stems from the research efforts of the Grimme group in Bonn and we
are indebted to their ingenious contributions to ORCA. Noticeably, there are a number of super-fast DFT
variants in ORCA 4.0 for ground- and excited state calculations that enable the user for unprecedented
throughput while maintaining essentially the full accuracy of the DFT methodology.
We have decided to now release ORCA 4.0 since there has been so much progress along many lines. However,
there are a significant number of developments that are close to be finalized. Hence we plan to release an
update (ORCA 4.1) in less than twelve months in order to let our users participate in the continuing progress
of the program.
If you, as an academic user, make use of ORCA and appreciate it, then PLEASE pay us back by referencing
the relevant original literature from our research group. This ensures that we can document the importance
of our developments and raise the funds that we need to continue the development. We sincerely hope that
with the release of ORCA 4.0 you will be able to push your research to the next level. We look very forward
to read about your results in the literature and discuss your insights together with other users in our forum
or to meet you at one of the next ORCA user meetings. If you are an ambitious method developer and feel
that you want to join the team, we look forward to your application!
Frank Neese, on behalf of the ORCA development team, march 1st, 2017
XVIII 1 Foreword 2017 (Version 4.0)
A final word to the academic users of this program package:
You can count on the fact that ORCA will remain free of charge in the future.
The best way to show your appreciation of our efforts is:
Cite our original research and development papers
in your own work!
This will provide us with the possibility of raising sufficient funding

for continuing to do what we enjoy most
develop quantum chemistry to our mutual benefit!
XIX
2 ORCA 4.0 Changes
As mentioned in the foreword, from ORCA 3.0.3 to ORCA 4.0 a lot has changed. Below is a summary of the
new features, which are thoroughly described in the respective sections of the manual. Furthermore, we list
changes in the default settings as well as bugfixes that will affect the computed results compared to older
version of ORCA.
2.1 New Features
New Methods
Linear scaling DLPNO-CCSD(T) open shell. New restricted open-shell formulation

Linear scaling DLPNO-MP2 (RHF and UHF)
Linear scaling DLPNO-MP2-F12 (RHF)
Linear scaling DLPNO-CCSD(T) (the 2013 implementation is still available)
Linear scaling DLPNO-CCSD(T) local energy decomposition scheme
Linear scaling DLPNO-CCSD closed shell density
Linear scaling cluster in molecule (CIM): MP2, CCSD(T), DLPNO-CCSD(T)
Linear scaling DLPNO-NEVPT2 (section 9.12)
NEVPT2-F12 (section 9.12)
Updated interface to BLOCK 1.0 (section 9.13)
DMRG-NEVPT2 (section 9.12)
Closed shell EOM-CCSD energies
Closed shell STEOM-CCSD energies
A partial PNO-EOM-CCSD method for excited states
A partial PNO-STEOM-CCSD method for excited states
Non linear scaling LPNO-CCSD-F12 (DLPNO-CCSD-F12 pending)
Non-linear scaling Mukherjee Mk-LPNO-MRCCSD(T)
Powerful iterative configuration expansion (ICE-CI) approximation to Full-CI (section 9.15)
ICE-CI for large active space CASSCF calculations (page 418)
MREOM-CCSD (also with SOC)
Fully internally contracted MRCI (section 9.16.3)
Full TD-DFT energies and gradient for hybrid functionals
Super-fast approximate TD-DFT: sTDA/sTDDFT of Grimme and co-workers
PBEh-3c method of Grimme and co-workers
SCF, DFT and Hessian
Large performance improvements (up to factor of four) for calculations with four center integrals
(energy and gradient)
XX 2 ORCA 4.0 Changes
Improved performance with RI-J with conventionally stored integrals

Gradient for range separated hybrids
Gradient for range double hybrid functionals with meta GGAs
Gradient for range double hybrid functionals with range separated functionals
Gradient for RI-JK
Frequencies for range separated functionals
Stability analysis and automatic search for broken symmetry states
Local spin analysis (defined via fragments in the coordinate block)
Fractional occupation number analysis (FOD) for detection of MR character
MDCI module
Automatic closed shell STEOM-CCSD active space selection

EOM-CCSD(2) and STEOM-CCSD(2) approximations
EOM-CCSD transition moments
EOM/STEOM-CCSD core level excited states
IP-EOM-CCSD and EA-EOM-CCSD
COSX for EOM-CCSD and STEOM-CCSD
AUTOCI module
RHF CISD
RHF CCSD
UHF CISD
UHF CCSD
ROHF CISD
ROHF CCSD
FIC-MRCI, CEPA/0 variant and DDCI3 (section 9.16.3)
CASSCF and NEVPT2
Accelerated CI (ACCCI) a more efficient CI step for multi-root calculations (page 418)
Automatic implementation of ab initio ligand-field theory (page 417)
Simplified generation of double-shell orbitals e.g. second d-shell (page 417)
Active space protection scheme and improved warnings (page 416)
ICE-CI as CI solver for larger active spaces (page 418)
New default converger (SuperCI PT) and convergence settings
Partially Contracted NEVPT2 with and without RI (section 9.12)
Updated interface to BLOCK 1.0 (section 9.13)
DMRG-NEVPT2 for active spaces up to 20 orbitals(section 9.12)
Magnetization and magnetic susceptibility (section 9.23.2.4)
Printing of the wavefunction in terms of CSFs and spin-determinants (page 417)
TD-DFT and ROCIS
Full TD-DFT for hybrid functionals

Gradient for full TD-DFT with hybrid functionals
TD-DFT/TDA gradient with range separated functionals
2.2 Changes of Default Settings & Parameters XXI
ROCIS MCD spectra

ROCIS-RIXS spectra
Natural transition orbitals in TD-DFT and ROCIS
Misc
New correlation consistent basis sets added

New SARC quadruple-zeta basis sets for the lanthanides
ANO-RCC basis sets added
new guess for ANO calculations (ehtano)
Sapporo-2012 basis sets added
Improved frozen core handling in correlation calculations (section 9.8)
Corrections for low-frequency modes in thermochemistry
New and improved NBO interface
New GEPOL surface construction (SES & SAS)
CPCM and improved SMD solvent models
Intrinsic atomic orbital (IAO) and bond orbital implementation (section 9.32.5)
Improved performance in Boys localization
Updated and improved orca mapspc program
New utility program orca exportbasis to print and store ORCA basis sets as files (section 9.32.12)
Atomic Mean Field (AMFI) spin-orbit coupling operators
EPRNMR works with range separated hybrid functionals
New AutoAux construction scheme (section 9.3)
2.2 Changes of Default Settings & Parameters

New handling of basis set names that is consistent with TurboMole def2-defaults using ECPs (section
3 and 9.3 ).
New reading of basis sets and ECPs together
New input handling for ANO basis sets
New frozencore definition across the periodic table (section 9.8).
Improved automatic frozen core handling
NEVPT2 uses frozencore by default (page 427)
Core-correlation in automatic basis set extrapolation
MRCI with CASSCF reference wavefunction now considers [ISSA] type excitations by default. As a
reminder excitation class are controlled by the keyword flags[ISSA] in the %mrci block.
CASSCF new default weighting for state-averaging (equal weights for multiplicity blocks). The old
default used equal weights for all states. Note that weights can still be set manually as described in
the CASSCF section.
2.3 Important Bugfixes with respect to ORCA 3.0.3

RIJCOSX SCF IR-Intensities (wrong numbers)
MP2 and RI-MP2 IR-Intensities (wrong numbers)
XXII 2 ORCA 4.0 Changes
Batching in the RI-MP2 Hessian(wrong numbers)

Batching RIJK conv in CASSCF (wrong numbers)
Some inconsistencies in the aug-cc-PCVDZ
Some inconsistencies in the ma-def2-QZVPP (Mn, Fe, Co, Ni, Cu, Zn)
Some inconsistencies in the def2-QZVPP (Mn, Fe, Co, Ni, Cu, Zn)
Added missing ECP contribution to the SCF Hessian (wrong numbers)
!UseSymmetry for D2 point group symmetry (B2/B3 mixed up)
MP2 and RI-MP2 Gradients with frozencore (slightly wrong numbers)
Reading of orbitals with different charge in SCF calculations (wrong number of electrons)
RI-CCSD energies (wrong numbers due to missing term). RI-CCSD(T) was not affected (OK).
Generating QROs in combination with !NoIter
Finite Field calculations with electric field as perturbation (sign convention)
SMD free energies for oxygen containing molecules in non-aqueous solutions (wrong numbers)
XXIII
3 FAQ frequently asked questions
Noticeable changes between ORCA 2.x, 3.x and ORCA 4.0
ORCA 4.0 exhibits a new basis sets convention.
From ORCA 4.0 onwards, the standard basis-sets contain ECPs
ECPs are specified using simple names, e.g. ! DEF2-ECP, the old ECP{...} is deprecated.
When using relativistic options, recontracted basis sets have to be given explicitly,
e.g. ! ZORA ZORA-DEF2-TZVP
Basis set names in the basis and coordinates block have to be specified in quotation marks.
Especially the conversion of old input files to the new ORCA 4.0 format has to be mentioned. To get the
same results with ORCA 4.0 as with previous ORCA versions the following rules apply.
Old (pre 4.0) New (> ORCA 4.0)

--------------------------------------------------------------------------------------------
! RKS BP RI ZORA DEF2-SVP DEF2-SVP/J ! RKS BP RI ZORA ZORA-DEF2-SVP DEF2/J
%basis basis CC_PVDZ %basis basis "CC-PVDZ"

aux CC_PVDZ_C aux "CC-PVDZ/C"
end end
*xyz 0 1 *xyz 0 1
... ...
* *
ORCA 4.0 exhibits new frozencore definitions across the periodic table, see section 9.8.
The standard computational levels (section 9.2.2.9) are deprecated and may no longer work as expected (in
particular, they will not assign a basis set).
Why is ORCA called ORCA?
Frank Neese made the decision to write a quantum chemistry program in the summer of 1999 while finishing
a postdoc at Stanford University. While thinking about a name for the program he wanted to write he
decided against having yet another whatever-Mol-something. The name needed to be short and signify
something strong yet elegant.
XXIV 3 FAQ frequently asked questions
During this time in the US Frank went on a whale watching cruise at the California coastthe name ORCA
stuck. It is often get asked whether ORCA is an acronym and over the years, various people made suggestions
what acronym this could possibly be. At the end of the day it just isnt an acronym which stands for anything.
It stands for itself and the association which comes with it.
How do I install ORCA on Linux / MacOS / Windows?
ORCA is available for Windows, Linux and Mac OS X platforms. A good place to start looking for detailed
installation instructions aside from the manual is the ORCA input library. Windows users furthermore have
the option of following this video description.
Ive installed ORCA, how do I start it?
First and most importantly, ORCA is invoked from the command line on all platforms. A simple click on
a binary or an input file wont start a calculation. Under Linux and MacOS you need to open a terminal
instance and navigate to the folder containing an example.inp file. You can run an ORCA calculation with
the command:
<full orca binary folder path>/orca example.inp > example.out
Similarly, under Windows you need to open a command prompt (Win7, Win8) or a power shell (Win10),
navigate to said directory and execute the following command:
<full orca binary folder path>/orca example.inp > example.out
How do I cite ORCA?
Please do NOT just cite the generic ORCA reference given below but also cite in addition our original
papers! We give this program away for free to the community and it is our pleasure and honour to do so. Our
payment are your citations! This will create the visibility and impact that we need to attract funding which
in turn allows us to continue the development. So, PLEASE, go the extra mile to look up and properly cite
the papers that report the development and ORCA implementation of the methods that you have used in
your studies!
The generic reference for ORCA is:

Neese, F.The ORCA program system Wiley interdisciplinary Reviews - Computational Molecular Science,
2012, Vol 2., Issue 1, Pages 7378
XXV
Are there recommended programmes to use alongside ORCA?
As a matter of fact there are: We make extensive use of Chemcraft. It is interesting to note that it works
well in MacOS or Linux (using Wine). Another popular visualization programme is Chimera.
OpenBabel is very useful for file conversion to various chemical formats.
Finally, Avogadro is an excellent tool to edit molecular geometries. It is also able to generate ORCA input
files. The Avogadro version with the latest ORCA modifications is available on the ORCA download site.
For other valuable suggestions please refer to the corresponding ORCA site.
My old inputs dont work with the new ORCA version! Why?
Please be aware that ORCA has changed considerably from 3-0 to 4-0. The basis set naming convention has
changed as well as a couple of defaults (frozen core e.g.). (For detailed information please refer to the Release
Notes.) It therefore is not unexpected that the same inputs will now either give slightly different results, or
will totally crash, because the keywords have been changed or totally removed.
If you are unsure about an input, please consult the manual. It is provided by the ORCA developers and
should contain all information implemented in the published version of ORCA.
When dealing with array structures, when does ORCA count starting from zero and
in which cases does counting start from one?
Since ORCA is a C++ based program its internal counting starts from zero. Therefore all electrons, atoms,
frequencies, orbitals, excitation energies etc. are counted from zero. User based counting such as the
numeration of fragments is counted from one.
How can I check that my SCF calculation converges to a correct electronic structure?

The expectation value S 2 is an estimation of the spin contamination in the system. It is highly recom-
mended for open-shell systems, especially with transition metal complexes, to check the UCO (unrestricted
corresponding orbitals) overlaps and visualise the corresponding orbitals. Additionally, spin-population on
atoms that contribute to the singly occupied orbitals is also an identifier of the electronic structure.
I cant locate the transition state (TS) for a reaction expected to feature a low/very
low barrier, what should I do?
For such critical case of locating the TS, running a very fine (e.g. 0.01
A increment of the bond length) relaxed
scan of the key reaction coordinate is recommended. In this way the highest energy point on a very shallow
surface can be identified and used for the final TS optimisation.
XXVI 3 FAQ frequently asked questions
During the geometry optimisation some atoms merge into each other and the
optimisation fails. How can this problem be solved?
This usually occurs due to the wrong or poor construction of initial molecular orbital involving some atoms.
Check the basis set definition on problematic atoms and then the corresponding MOs!
While using MOREAD feature in ORCA, why am I getting an error saying, no orbitals
found in the .gbw file?
ORCA produces the .gbw file immediately after it reads the coordinates and basis set information. If you put
a .gbw file from a previous calculation with same base name as your current input into the working directory,
it will be overwritten and the previous orbital data will be lost. Therefore, it is recommended to change the
file name or .gbw extension to something else (.gbw.old, for example).
The localisation input file (.loc.fil) I used to use for older ORCA versions is not
working with orca loc of ORCA 4.0.
The content of the localization input file has been modified in the current ORCA version. Now some additional
input data is required for localisation. Type orca loc in a shell and you will get the list of required input
information.
With all the GRID and RI and associated basis set settings Im getting slightly
confused. Can you provide a brief overview?
HartreeFock (HF) and DFT require the calculation of Coulomb and exchange integrals. While the Coulomb
integrals are usually done analytically, the exchange integrals can be evaluated semi-numerically on a grid.
Here, the pure DFT exchange is calculated on one type of grid (controlled through the GRID keyword)
while the HF exchange can be evaluated on an different, often smaller grid (GRIDX). For both parts, further
approximations can be made (RI-J and RI-K1 or COSX, respectively). When RI is used, axillary basis sets
are required (<basis>/ J for RI-J and <basis>/ JK for RI-JK). The following possible combinations arise:
HF calculation
Exact J + exact K: no auxiliary functions and no grids needed.

RIJ + exact K (RIJONX, RIJDX): <basis>/ J auxiliaries, no grids.
RIJ + RIK = RIJK: <basis>/JK auxiliaries, no grids.
RIJ + COSX: <basis>/ J auxiliaries, COSX grid controlled by the GRIDX keyword.
GGA DFT functional
Exact J + GGA-XC: no auxiliary functions needed, DFT grid controlled by the GRID keyword.
RIJ + GGA-XC: <basis>/ J auxiliaries, DFT grid controlled by the GRID keyword.
Hybrid DFT functional
Exact J + exact K + GGA-XC: no auxiliary functions needed, DFT grid controlled by the GRID keyword.
1
Note that ORCA can only use RI-K in conjunction with RI-J; hence the combination RI-JK.
XXVII
RIJ + exact K (RIJONX, RIJDX) + GGA-XC: <basis>/ J auxiliaries, DFT grid controlled by the GRID
keyword.
RIJ + RIK (RIJK) + GGA-XC: <basis>/ JK auxiliaries, DFT grid controlled by the GRID keyword.
RIJ + COSX + GGA-XC: <basis>/ J auxiliaries, COSX grid controlled by the GRIDX keyword, DFT
grid controlled by the GRID keyword.
There are a lot of basis sets! Which basis should I use when?
ORCA offers a variety of methods and a large choice of basis sets to go with them. Here is an incomplete
overview:
Method Approximation basis set (and auxiliaries)
CASSCF/NEVPT2 <basis>
CASSCF/NEVPT2 RIJCOSX or RI-JK <basis>+ <basis>/JK
CASSCF/NEVPT2 TrafoStep RI <basis>+ <basis>/JK or <basis>/C
NEVPT2-F12 TrafoStep RI <basis>-F12 + <basis>-F12/CABS + <basis>/JK or <basis>/C
TDDFT <basis>
TDDFT Mode RIInts <basis>+ <basis>/C
MP2 <basis>
F12-MP2 <basis>-F12 + <basis>-F12/CABS
RI-MP2 <basis>+ <basis>/C
HF+RI-MP2 RIJCOSX <basis>+ <basis>/C + <basis>/J
F12-RI-MP2 <basis>-F12 + <basis>-F12/CABS + <basis>/C
DLPNO-MP2 <basis>+ <basis>/C
HF+DLPNO-MP2 RI-JK <basis>+ <basis>/C + <basis>/JK
F12-DLPNO-MP2 <basis>-F12 + <basis>-F12/CABS + <basis>/C
CCSD <basis>
RI-CCSD <basis>+ <basis>/C
(D)LPNO-CCSD <basis>+ <basis>/C
HF+(D)LPNO-CCSD RIJCOSX <basis>+ <basis>/C + <basis>/J
F12-CCSD <basis>-F12 + <basis>-F12/CABS
F12-RI-CCSD <basis>-F12 + <basis>-F12/CABS + <basis>/C
HF+F12-RI-CCSD RI-JK <basis>-F12 + <basis>-F12/CABS + <basis>/C + <basis>/JK
XXVIII
1
4 General Information
4.1 Program Components
The program system ORCA consists of several separate programs that call each other during a run. The
following basic modules are included in this release:
orca : Main input+driver program

orca anoint : Integral generation over ANOs
orca autoci : CI type program using the automated generation environment (ORCA-AGE)
orca ciprep : Preparation of data for MRCI calculations (frozen core matrices and the like)
orca cis : Excited states via CIS and TD-DFT
orca cipsi : Iterative Configuration Expansion Configuration Interaction (ICE-CI)
orca cpscf : Solution of the coupled-perturbed SCF equations
orca casscf : Main program for CASSCF driver
orca eprnmr : SCF approximation to EPR and NMR parameters
orca fci : Full-CI program
orca gtoint : Calculation of gaussian integrals
orca gstep : Relaxation of the geometry based on energies and gradients
orca loc : Calculation of localized molecular orbitals
orca md : Molecular dynamics program
orca mdci : Matrix driven correlation program: CI, CEPA, CPF, QCISD, CCSD(T)
orca mp2 : MP2 program (conventional, direct and RI)
orca mrci : MRCI and MRPT calculations (individually selecting)
orca ndoint : Calculates semiempirical integrals and gradients
orca numfreq : Numerical hessian computation
orca pc : Addition of point charge terms to the one-electron matrix
orca plot : Generation of orbital and density plots
orca pop : External program for population analysis on a given density
orca rel : (Quasi) Relativistic corrections
orca rocis : Excited states via the ROCIS method
orca scf : Self-consistent field program (conventional and direct)
orca scfgrad : Analytic derivatives of SCF energies (HF and DFT)
orca scfhess : Analytical hessian calculation for SCF
orca soc : Calculation of spin-orbit coupling matrices
orca vpot : Calculation of the electrostatic potential on a given molecular surface
Utility programs:
2 4 General Information
orca asa : Calculation of absorption, fluorescence and resonance Ra-

man spectra
orca vib : Calculation of vibrational frequencies from a completed fre-
quency run (also used for isotope shift calculations)
orca pltvib : Produces files for the animation of vibrations
orca 2mkl : Produces an ASCII file to be read by molekel, molden or
other visualization programs
orca 2aim : Produces WFN files suitable for AIM analysis
orca fitpes : Simple program to fit potential energy curves of diatomics
orca mapspc : Produces files for transfer into plotting programs
orca euler : Calculate Euler angles from .prop file
orca chelpg : Electrostatic potential derived charges
otool gcp : Geometrical Counterpose Correction
Friends of ORCA:
gennbo : The NBO analysis package of Weinhold (must be purchased

separately from the university of Wisconsin; older versions
available for free on the internet may also work)
Molekel : Molecular visualization program (see 7.20.2.3)
gOpenMol : Molecular visualization program (see 7.20.2.2)
In principle every individual module can also be called standalone. However, it is most convenient to do
everything via the main module.
There is no real installation procedure. Just copy the executables wherever you want them to be and make
sure that your path variable contains a reference to the directory where you copied the files. This is important
to make sure that the programs can call each other (but you can also tell the main program the explicit
position of the other programs in the input file as described below).
4.2 Units and Conversion Factors
Internally the program uses atomic units. This means that the unit of energy is the Hartree (Eh) and the
unit of length is the Bohr radius (a0 ). The following conversion factors to other units are used:
1 Eh = 27.2113834 eV
1 eV = 8065.54477 cm1 = 23.0605 kcal
mol
1 cm1 = 29979.2458 MHz
1 a0 = 0.5291772083
A
1 a.t.u. = 2.4188843 1017 s
3
5 Calling the Program (Serial and Parallel)
5.1 Calling the Programm
Under Windows the program is called from the command prompt! (Make sure that the PATH variable is set
such that the orca executables are visible)
orca MyMol.inp > MyMol.out
Under UNIX based operating systems the following call is convenient1 (here also: make sure that the PATH
variable is set to the directory where the orca executables reside):
nohup orca MyMol.inp >& MyMol.out &
The nohup command lets the program run even if the user is logged out. The program writes to stdout
and stderr. Therefore the output must be redirected to the file MyMol.out in this example. MyMol.inp is
a free format ASCII file that contains the input description. The program will produce a number of files
MyMol.x.tmp and the file MyMol.gbw. The *.gbw file contains a binary summary of the calculation. GBW
stands for Geometry-Basis-Wavefunction. Basically this together with the calculation flags is what is stored
in this file. You need this file for restarting SCF calculations or starting other calculations with the orbitals
from this calculation as input. The *.tmp files are temporary files that contain integrals, density matrices,
Fock matrices etc. that are used as intermediates in the calculation. If the program exits normally all of
these files are deleted. If it happens to crash you have to remove the files manually (rm MyMol*.tmp under
Unix or del MyMol*.tmp under Windows). In case you want to monitor the output file while it is written,
you can use the command (under Unix):
tail -f MyMol.out
to follow (option -f) the progress of the calculation. Under Windows you have to either open another command
shell and use:
type MyMol.out
type MyMol.out |more
1
Many people (including myself) will prefer to write a small shellscript that, for example, creates a run directory,
copies the input there, runs the program, deletes possibly left over temporary files and then copies the output back
to the original directory.
4 5 Calling the Program (Serial and Parallel)
or you have to copy the output file to another file and then use any text editor to look at it.
copy MyMol.out temp.out

edit temp.out
you cannot use edit MyMol.out because this would result in a sharing violation.
There are parallel versions for Linux, MAC and Windows computers (thanks to the work of Ms Ute Becker)
which make use of OpenMPI (open-source MPI-2 implementation). Assuming that OpenMPI libraries are
installed properly on your computer, it is fairly easy to run the parallel version of ORCA. You simply have
to specify the number of parallel processes, like:
! PAL4
# everything from PAL2 to PAL8 is recognized
or
%pal nprocs 4 # any number (positive integer)

end
The following modules are presently parallelized:
ANOINT
CASSCF / NEVPT2
CIPSI
CIS/TDDFT
CPSCF
EPRNMR
GTOINT
MDCI (Canonical-, PNO-, DLPNO-Methods)
MP2 and RI-MP2 (including Gradient and Hessian)
MRCI
PC
ROCIS
SCF
SCFGRAD
SCFHESS
5.2 Hints on the Use of Parallel ORCA 5
SOC
Numerical Gradients and Frequencies
Thus, all major modules are parallelized in the present version. The efficiency is such that for RI-DFT
perhaps up to 16 processors are a good idea while for hybrid DFT and Hartree-Fock a few more processors are
appropriate. Above this, the overhead becomes significant and the parallelization loses efficiency. Coupled-
cluster calculations usually scale well up to at least 8 processors but probably it is also worthwhile to try 16.
For Numerical Frequencies or Gradient runs it makes sense to use as many processors as 3*Number of Atoms.
If you run a queuing system you have to make sure that it works together with ORCA in a reasonable way.
NOTE:
Parallelization is a difficult undertaking and there are many different protocols that work differently
for different machines. Please understand that we can not provide a 1:1 support for each platform.
We are trying our best to make the parallelization transparent and provide executables for various
platforms but we can not possibly guarantee that they always work on every system. Please see the
download information for details of the version.
5.2 Hints on the Use of Parallel ORCA
Many questions that are asked in the discussion forum deal with the parallel version of ORCA. Please
understand that we cannot possibly provide one-on-one support for every parallel computer in the world. So
please, make every effort to solve the problems locally together with your system administrator. Here are
some of the most common problems and how to deal with them.
1. Parallel ORCA can be used with OpenMPI only. Please see the download information for details of the
relevant OpenMPI-version for your platform.
2. The OpenMPI version is configurable in a large variety of ways, which cannot be covered here. For a more
detailed explanation of all available options, cf. http://www.open-mpi.org
Otherwise, the usage of the OpenMPI version is like the older MPICH version, especially with regard to the
provision of a nodefile (<inputfile>.nodes). If you run the OpenMPI version on only one computer, you do
not need to provide a machinefile, and neither have to enable an rsh/ssh access, as in this case the processes
will simply be forked!
Please note that the OpenMPI version is dynamically linked, that is, it needs at runtime the OpenMPI
libraries (and several other standard libraries)!
(Remember to set the LD LIBRARY PATH)
3. Many problems arise, because parallel ORCA does not find its executables. To avoid this, it is crucial to
provide ORCA with its complete pathname. The easiest and safest way to do so is to include the directory
with the orca-executables in your $PATH. Then start the calculation:
- interactively: start orca with full path: /mypath orca executables/orca MyMol.inp
- batch : export your path: export PATH=$PATH:/mypath orca executables (for bash) then start orca with
full path: $PATH/orca $jobname.inp
This seems redundant, but it really is important if you want to start a parallel calculation to run ORCA
with full path! Otherwise it will not be able to find the parallel executables.
4. It is recommended to run orca in local (not nfs-mounted) scratch-directories, (for example /tmp1, /usr/local,
...) and to renew these directories for each run to avoid confusion with left-overs of a previous run.
5. It has proven convenient to use wrapper scripts. These scripts should
set the path
create local scratch directories
copy input files to the scratch directory
start orca
save your results
remove the scratch directory
A basic example of such a submission script for the parallel ORCA version is shown below (this is for the
Torque/PBS queuing system, running on Apple Mac OS X):
#!/bin/zsh
setopt EXTENDED_GLOB
setopt NULL_GLOB
#export MKL_NUM_THREADS=1
b=${1:r}
#get number of procs.... close your eyes... (it really works!)

if [[ ${$(grep -e ! $1):u} == !*(#b)PAL(<0-9>##)* ]]; then
nprocs=$match
let "nodes=nprocs"
elif [[ ${(j: :)$(grep -v # $1):u} == *%(#b)PAL*NPROCS #(<0-9>##)* ]]; then
nprocs=$match
let "nodes=nprocs"
fi
cat > ${b}.job <<EOF

#!/bin/zsh
#PBS -l nodes=1:ppn=${nodes:=1}
#PBS -S /bin/zsh
#PBS -l walltime=8760:00:00
setopt EXTENDED_GLOB
setopt NULL_GLOB
export PATH=$PBS_O_PATH
logfile=$PBS_O_WORKDIR/$b.log
5.2 Hints on the Use of Parallel ORCA 7
tdir=$(mktemp -d /Volumes/scratch/$USER/$b__XXXXXX)
trap
echo "Job terminated from outer space!" >> $logfile
rm -rf $tdir
exit
TERM
cp $PBS_O_WORKDIR/$1 $tdir
foreach f ($PBS_O_WORKDIR/*.gbw $PBS_O_WORKDIR/*.pot) cp $f $tdir
cd $tdir
echo "Job started from $PBS_O_HOST, running on $(hostname) in $tdir using

$(which orca)" > $log
file
=orca $1 1>>$logfile 2>&1
cp (*.(inp|tmp*)) $PBS_O_WORKDIR/
rm -rf $tdir
EOF
qsub -j oe -o $b.job.out $b.job
6. Parallel ORCA distinguishes 3 cases of disk availability:
each process works on its own (private) scratch directory (the data on this directory cannot be seen
from any other process)
all processes work in a common scratch directory (all processes can see all file-data)
there are at least 2 groups of processes on different scratch directories, one of the groups consisting of
more than 1 process
Parallel ORCA will find out, which case exists and handle its I/O respectively. If ORCA states disk availability
differently from what you would expect, check the number of available nodes and/or the distribution pattern
(fill up/round robin)
7. If Parallel ORCA finds a file named MyMol.nodes in the directory where its running, it will use the
nodes listed in this file to start the processes on, provided your input file was MyMol.inp. You can use this
file as your machinefile specifying your nodes, using the usual OpenMPI machinefile notation.
8. It is possible to pass additional MPI-parameters to the mpirun by adding these arguments to the ORCA
call:
/mypath orca executables/orca MyMol.inp --bind-to core
- or - for multiple arguments
/mypath orca executables/orca MyMol.inp "--bind-to core --verbose"

9. An additional remark on multi-process numerical calculations (frequencies, gradient, hybrid hessian):

The processes that execute these calculations do not work in parallel, but independently, often in a totally
asynchronous manner. The numerical calculations will start as many processes, as you dedicated for the
parallel parts before and they will run on the same nodes. If your calculation runs on multiple nodes, you have
to set the environment variable RSH COMMAND to either rsh or ssh. You may specify special flags, like
ssh -x. If RSH COMMAND is not set, ORCA will start all processes of a multi-process run on localhost.
(Take care not to exceed your localhosts ressources!) There is no gain in taking more processes than 3-times
the number of atoms to be displaced. For restart (available for numerical frequencies, hybrid hessian and
partial hessian calculations) make sure you have all local Hessian files (input.proc%d.hess). ORCA will check
these files to determine which displacements are left to be done.
9
6 General Structure of the Input File
In general, the input file is a free format ASCII file and can contain one or more keyword lines that start
with a ! sign, one or more input blocks enclosed between an % sign and end that provide finer control
over specific aspects of the calculation, and finally the specification of the coordinates for the system along
with the charge and multiplicity provided either with a %coords block, or more usually enclosed within two
* symbols. Here is an example of a simple input file that contains all three input elements:
! HF def2-TZVP
%scf
convergence tight
end
* xyz 0 1
C 0.0 0.0 0.0
O 0.0 0.0 1.13
*
Comments in the file start by a #. For example:
# This is a comment. Continues until the end of the line
Comments can also be closed by a second #, as the example below where TolE and TolMaxP are two
variables that can be user specified:
TolE=1e-5; #Energy conv.# TolMaxP=1e-6; #Density conv.#
The input may contain several blocks, which consist of logically related data that can be user controlled. The
program tries to choose sensible default values for all of these variables. However, it is impossible to give
defaults that are equally sensible for all systems. In general the defaults are slightly on the conservative
side and more aggressive cutoffs etc. can be chosen by the user and may help to speed things up for actual
systems or give higher accuracy if desired.
10 6 General Structure of the Input File
6.1 Input Blocks
The following blocks exist:
Autoci Controls autogenerated correlation calculations

Basis Basis sets are specified
Casscf Control of CASSCF/NEVPT2 and DMRG calculations
Cipsi Control of Iterative-Configuration Expansion Configuration Interaction calcula-
tion
Cis Control of CIS and TD-DFT calculations (synonym is TDDFT)
Cim Control of Cluster In Molecules calculations
Coords Input of atomic coordinates
Cpcm Control of the Conductor-like Polarizable Continuum Model
Elprop Control of electric property calculations
Eprnmr Control of SCF level EPR and NMR calculations
Freq Control of frequency calculations
Geom Control of geometry optimization
Loc Localization of orbitals
Md Control of molecular dynamics simulation

Mdci Controls single reference correlation methods
Method Here a computation method is specified
Mp2 Controls the details of the MP2 calculation
Mrci Control of MRCI calculations
Mrcc Control of multi-reference CC calculations
Numgrad Control of numerical gradients
Nbo Controls NBO analysis with GENNBO
Output Control of output
Pal Control of parallel jobs
Paras Input of semi-empirical parameters
Rel Control of relativistic options
Plots Control of plot generation
Rocis Control of restricted-open-shell CIS
Rr Control of resonance Raman and absorption/fluorescence band-shape calculations
Scf Control of the SCF procedure
Vpt2 Control of the anharmonic frequency calculations
Blocks start with % and end with end. Note that input is not case sensitive. For example:
%method method HF
end
No blocks need to be present in an input file but they can be present if detailed control over the behavior of
the program is desired. Otherwise all normal jobs can be defined via the keywords described in the next
section. Variable assignments have the following general structure:
6.2 Keyword Lines 11
VariableName Value
Some variables are actually arrays. In this case several possible assignments are useful:
Array[1] Value1
Array[1] Value1,Value2,Value3
Array Value1,Value2
Note: Arrays always start with index 0 in ORCA (this is because ORCA is a C++ program). The first line in
the example gives the value Value1 to Array[1], which is the second member of this array. The second line
assigns Value1 to Array[1], Value2 to Array[2] and Value3 to Array[3]. The third line assigns Value1
to Array[0] and Value2 to Array[1]. Strings (for examples filenames) must be enclosed in quotes. For
example:
MOInp "Myfile.gbw"
In general the input is not case sensitive. However, inside strings the input is case sensitive. This is because
on unix systems MYFILE.GBW and MyFile.gbw are different files. Under Windows the file names are not case
sensitive.
6.2 Keyword Lines
It is possible to give a line of keywords that assign certain variables that normally belong to different input
blocks. The syntax for this simple input is line-oriented. A keyword line starts with the ! sign.
! Keywords
6.2.1 Main Methods and Options
Table 6.1 provides a list of keywords that can be used within the simple input keyword line to request
specific methods and/or algorithmic options. Most of them are self-explanatory. The others are explained in
detail in the section of the manual that deals with the indicated input block.
Table 6.1: Main keywords that can be used in the simple input of ORCA.
Keyword Input Variable Comment
block
HF METHOD METHOD Selects the HartreeFock method
DFT Selects the DFT method (see table 6.2
on page 22 for a list of functionals)
FOD FOD analysis (see 9.5.8.2) employ-
ing default settings (TPSS/def2-TZVP,
TightSCF, SmearTemp = 5000 K)
Runtypes
ENERGY or SP METHOD RUNTYP Selects a single point calculation
OPT Selects a geometry optimization calcu-
lation (using internal redundant coordi-
nates)
COPT Optimization in Cartesian coordinates
(if you are desperate)
ZOPT Optimization in Z-matrix coordinates
(dangerous)
GDIIS-COPT COPT using GDIIS
GDIIS-ZOPT ZOPT using GDIIS
GDIIS-OPT Normal optimization using GDIIS
ENGRAD Selects an energy and gradient calcula-
tion
NUMGRAD Numerical gradient (has explicitly to
be asked for, if analytic gradient is not
available)
NUMFREQ Numerical frequencies
MD Molecular dynamic simulation
CIM Cluster-In-Molecule calculation
Atomic mass/weight handling
Mass2016 METHOD AMASS Use the latest (2016) atomic masses
of the most abundant or most stable
isotopes instead of atomic weights (de-
fault).
Symmetry handling
UseSym Turns on the use of molecular symme-
try (see section 6.4). THIS IS VERY
RUDIMENTARY!
NoUseSym Turns symmetry off
Second order many body perturbation theory
MP2 Selects Method=HF and DoMP2=true
MP2RI or RI-MP2 Select the MP2-RI method
SCS-MP2 Spin-component scaled MP2
RI-SCS-MP2 Spin-component scaled RI-MP2
(synonym is SCS-RI-MP2)
OO-RI-MP2 Orbital optimized RI-MP2
OO-RI-SCS-MP2 Orbital optimized and spin-component

scaled RI-MP2
MP2-F12 MP2 with F12 correction
(synonym is F12-MP2)
MP2-F12-RI MP2-RI with RI-F12 correction
(synonyms are F12-RI-MP2, RI-MP2-
F12)
High-level single reference methods. These are implemented in the MDCI module.
They can be run in a number of technical variants.
CCSD MDCI CITYPE Coupled-cluster singles and doubles
CCSD(T) Same with perturbative triples correc-
tion
CCSD-F12 CCSD with F12 correction
CCSD(T)-F12 CCSD(T) with F12 correction
CCSD-F12/RI CCSD with RI-F12 correction
CCSD(T)-F12/RI CCSD(T) with RI-F12 correction
QCISD Quadratic Configuration interaction
QCISD(T) Same with perturbative triples correc-
tion
QCISD-F12 QCISD with F12 correction
QCISD(T)-F12 QCISD(T) with F12 correction
QCISD-F12/RI QCISD with RI-F12 correction
QCISD(T)-F12/RI QCISD(T) with RI-F12 correction
CPF/1 Coupled-pair functional
NCPF/1 A new modified coupled-pair func-
tional
CEPA/1 Coupled-electron-pair approximation
NCEPA/1 The CEPA analogue of NCPF/1
RI-CEPA/1-F12 RI-CEPA with F12 correction
MP3 MP3 energies
SCS-MP3 Grimmes refined version of MP3
Other coupled-pair methods are available and are documented later in the manual in detail
(section 7.8) In general you can augment the method with RI-METHOD in order to make
the density fitting approximation operative; RI34-METHOD does the same but only for the
3- and 4-external integrals). MO-METHOD performs a full four index transformation and
AO-METHOD computes the 3- and 4-external contributions on the fly. With AOX-METHOD
this is is done from stored AO integrals.
Local correlation methods. These are local, pair natural orbital based correlation
methods. They must be used together with auxiliary correlation fitting basis sets. Open-shell
variants are available for some of the methods, for full list please see section 6.1.3. We
recommend n = 1 for the CEPA methods.
LPNO-CEPA/n MDCI Various Local pair natural orbital CEPA meth-
ods
LPNO-CPF/n Same for coupled-pair functionals
LPNO-NCEPA/n Same for modified versions
LPNO-NCPF/n Same for modified versions
LPNO-QCISD Same for quadratic CI with singles and

doubles
LPNO-CCSD Same for coupled-cluster theory with
single and double excitations
DLPNO-CCSD Domain based local pair natural orbital
coupled-cluster method with single and
double excitations (closed-shell only)
DLPNO-CCSD(T) DLPNO-CCSD with perturbative triple
excitations (closed-shell only)
DLPNO-MP2 MP2 Various Local (DLPNO) MP2
DLPNO-SCS-MP2 Spin-component scaled DLPNO-MP2
(a synonym is SCS-DLPNO-MP2)
DLPNO-MP2-F12 DLPNO-MP2 with F12 correction
DLPNO-MP2- DLPNO-MP2-F12 with approach D
F12/D
DLPNO-NEVPT2 DLPNO-NEVPT2 requires a CASSCF
block
Accuracy control for local correlation methods. These keywords select predefined sen-
sible sets of thresholds to control the accuracy of DLPNO calculations. See the corresponding
sections on local correlation methods for more details.
LoosePNO MDCI, MP2 Various Selects loose DLPNO thresholds
NormalPNO Selects default DLPNO thresholds
TightPNO Selects tight DLPNO thresholds
Automatic basis set eaxtrapolation
Extrapolate (n/m, Extrapolation of the basis set fam-
bas) ily bas (bas=cc,aug-cc, cc-core, ano,
saug-ano, aug-ano, def2; if omitted
cc-pVnZ is used) for cardinal num-
bers n,m (n<m=2,3,4,5), e.g. Extrapo-
late(2/3,cc) extrapolates the SCF, MP2
and MDCI energies to the basis set
limit. core refers to basis sets with
core correlation function. In this case
the frozen core approximation is - by
default - turned off. This setting can
be overridden in the methods block if
one just wants to use the basis set with
core correlation functions (steep primi-
tives) but without unfreezing the core
electrons.
Extrapolate (n, Calculate the first n-energies for mem-
bas) ber of the basis set family basis, e.g. Ex-
trapolate(3) is doing calculations with
cc-pVDZ, cc-pVTZ and cc-pVQZ.
ExtrapolateEP2 Similar: performs SCF, MP2 and MDCI

(n/m, calculations. The higher basis set can
bas,[method,method- only be done with DLPNO-CCSD(T) or
details]) MP2 methods and then used to extrap-
olate the MDCI calculation to the basis
set limit.
ExtrapolateEP3 Similar to EP2: for the high basis set
(bas,[method,method- method we go one cardinal number
details]) higher.
CASSCF related options. All of them require the CASSCF block as minimal input
DMRG Sets DMRG as CIStep in CASSCF
NEVPT2 SC NEVPT2
SC-NEVPT2 SC-NEVPT2 same as NEVPT2
RI-NEVPT2 SC-NEVPT2 with the RI approximation
FIC-NEVPT2 FIC-NEVPT2 aka PC-NEVPT2
DLPNO-NEVPT2 FIC-NEVPT2 in the framework of
DLPNO
(internally contracted)Multireference methods beyond NEVPT2: If specified in a
single keyword all information about reference spaces, number of roots etc. is taken from the
CASSCF module that is assumed to be run in advance. These methods reside in the autoci
module. More refined settings require the autoci block in the input.
FIC-MRCI CIType Invokes the fully internally contracted
MRCI
FIC-CEPA0 Fully internally contracted CEPA0
(uncontracted)Multireference methods: If specified in a single keyword all information
about reference spaces, number of roots etc. is taken from the CASSCF module that is assumed
to be run in advance. In general, these calculations are of the individually selecting type and are
very time consuming. Very many flags can be set and modified for these methods and in general
using these methods requires expert users! In general see the variables Tsel, Tpre and Tnat
that define the individual selection process. All of these methods can be used with RI integrals
by using RI-MRCI etc. However, then the calculations become even more time consuming since
integrals are made one- by one on the fly. Non-RI calculations will be pretty much limited to
about 200-300 orbitals that are included in the CI
MRCI MRCI CIType Initiates a multireference configuration
interaction calculation with single and
double excitations
MRCI+Q Same with multireference Davidson cor-
rection for unlinked quadruples
MRACPF Average coupled-pair functional
MRAQCC Average quadratic coupled-cluster
MRDDCI1 Difference dedicated CI with one degree
of freedom
MRDDCI2 Same with two degrees of freedom
MRDDCI3 Same with three degrees of freedom
MRDDCIn+Q MRDDCI with Davidson correction
SORCI Spectroscopy oriented CI

Frozen core features. NOTE: this deviates from previous versions of ORCA! We are now
counting core electrons rather than using an energy window. If you do want to use an orbital
energy window use %method FrozenCore FC EWIN end. Otherwise the EWin commands will
be ignored! (alternatives are FC ELECTRONS (default) and FC NONE).
FROZENCORE METHOD FrozenCore Use a frozen core. By default this is done
by counting the number of chemical core
electrons
NOFROZENCORE Do not use a frozen core
Semiempirical methods
ZINDO/S Selects the ZINDO/S method
ZINDO/1 Selects the ZINDO/1 method
ZINDO/2 Selects the ZINDO/2 method
NDDO/1 Selects the NDDO/1 method
NDDO/2 Selects the NDDO/2 method
MNDO Selects the MNDO method
AM1 Selects the AM1 method
PM3 Selects the PM3 method
Algorithmic variations, options, add-ons, modifiers,. . .
RHF or RKS SCF HFTYP Selects closed-shell SCF
UHF or UKS Selects spin unrestricted SCF
ROHF or ROKS Selects open-shell spin restricted SCF
AllowRHF METHOD ALLOWRHF Allow a RHF calculation even if the sys-
tem is open-shell (Mult>1). Default is
to switch to UHF then
RI METHOD RI Sets RI=true to use the RI approxima-
tion in DFT calculations. Default to
Split-RI-J
NORI Sets RI=false
RIJCOSX METHOD/ RI, KMatrix Sets the flag for the efficient RIJCOSX
SCF algorithm (treat the Coulomb term via
RI and the Exchange term via seminu-
merical integration)
RI-JK METHOD/ RI, KMatrix Sets the flag for the efficient RI al-
SCF gorithm for Coulomb and Exchange.
Works for SCF (HF/DFT) energies (not
gradients!). Works direct or conven-
tional.
SPLITJ SCF JMATRIX Select the efficient Split-J procedure for
the calculation of the Coulomb matrix
in non-hybrid DFT (rarely used)
SPLIT-RI-J SCF JMATRIX,RI Select the efficient Split-RI-J procedure
for the improved evaluation of the RI-
approximation to the Coulomb-matrix
NoSplit-RI-J SCF JMATRIX,RI Turns the Split-RI-J feature off (but

does not set the RI flag to false!)
RI-J-XC SCF JMATRIX, KMA- Turn on RI for the Coulomb term and
TRIX,RI the XC terms. This saves time when the
XC integration is significant but intro-
duces another basis set incompleteness
error. (rarely used)
DIRECT SCF SCFMODE Selects an integral direct calculation
SEMIDIRECT Selects an integral semidirect calculation
(rarely used nowadays)
CONV Selects an integral conventional calcula-
tion
NOITER SCF MAXITER Sets the number of SCF iterations to
0. This works together with MOREAD
and means that the program will work
with the provided starting orbitals.
Initial guess options: In most cases the default PMODEL guess will be adequate. In some
special situations you may want to switch to a different choice
PATOM SCF GUESS Selects the polarized atoms guess
PMODEL Selects the model potential guess
HUECKEL Selects the extended H uckel guess
HCORE Selects the one-electron matrix guess
MOREAD Read MOs from a previous calulation
(use %moinp "myorbitals.gbw" in a
separate line to specify the GBW file
that contains these MOs to be read)
AUTOSTART AUTOSTART Try to start from the existing GBW file
of the same name as the present one
NOAUTOSTART Dont try to do that
Basis-set related keywords
Decontract BASIS Decontract the basis set. If the basis set
arises from general contraction, dupli-
cate primitives will be removed.
NoDecontract Prevent decontraction
DecontractAux Decontract the auxiliary basis set
NoDecontractAux Prevent decontraction of the auxiliary
basis
Relativistic options: There are several variants of scalar relativistic Hamiltonians to use
in all electron calculations
DKH or DKH2 REL METHOD/ORDER Selects the scalar relativistic Douglas
Kroll Hess Hamiltonian of 2nd order
ZORA REL METHOD Selects the scalar relativistic ZORA
Hamiltonian
ZORA/RI REL METHOD Selects the scalar relativistic ZORA
Hamiltonian in RI approximation
IORA/RI REL METHOD Selects the scalar relativistic IORA

Hamiltonian in RI approximation
IORAmm/RI REL METHOD Selects the scalar relativistic IORA mm
(modified metric) Hamiltonian in RI ap-
proximation
Grid options
GRIDn (n = 07) METHOD GRID Selects the DFT integration grid no n
FINALGRIDn Selects the DFT final integration grid
no n
NOFINALGRID Turns the final grid feature off
GRIDXn (n = 19) METHOD GRIDX Grids for the COSX approximation. A
sequence of three grids is used. Higher
accuracy at higher cost is offered by the
higher grids.
NOFINALGRIDX Turn off the final grid in COSX (not
recommended)
Convergence thresholds: These keywords control how tightly the SCF and geometry
optimizations will be converged. The program makes an effort to set the convergence
thresholds for correlation modules consistently with that of the SCF.
NORMALSCF SCF CONVERGENCE Selects normal SCF convergence
LOOSESCF Selects loose SCF convergence
SLOPPYSCF Selects sloppy SCF convergence
STRONGSCF Selects strong SCF convergence
TIGHTSCF Selects tight SCF convergence
VERYTIGHTSCF Selects very tight SCF convergence
EXTREMESCF Selects extreme convergence. All
thresholds are practically reduced to nu-
merical precision of the computer. Only
for benchmarking (very expensive).
SCFCONVn Selects energy convergence check and
sets ET ol to 10n (n = 610). Also se-
lects appropriate thresh, tcut, and bfcut
values.
TIGHTOPT GEOM TolE,TolRMSG Selects tight optimization convergence
NORMALOPT TolMaxG Selects default optimization convergence
LOOSEOPT TolRMSD,TolMaxD Selects loose optimization convergence
Convergence acceleration: the default is DIIS which is robust. For most closed-shell organic
molecules SOSCF converges somewhat better and might be a good idea to use. For trailing
convergence, KDIIS or the true second-order procedures NRSCF and AHSCF might be good
choices.
DIIS SCF DIIS Turns DIIS on
NODIIS Turns DIIS off
KDIIS SCF KDIIS Turns Kollmars DIIS on
NRSCF SCF NR Turns Newton-Raphson SCF on
AHSCF Same but with augmented Hessian step
NONRSCF Turns Newton-Raphson SCF off
SOSCF SCF SOSCF Turns SOSCF on

NOSOSCF Turns SOSCF off
DAMP SCF CNVDAMP Turns damping on
NODAMP Turns damping off
LSHIFT SCF CNVSHIFT Turns level shifting on
NOLSHIFT Turns level shifting off
Convergence strategies (does not modify the convergence criteria)
EasyConv Assumes no convergence problems.
NormalConv Normal convergence criteria.
SlowConv Selects appropriate SCF converger cri-
teria for difficult cases. Most transition
metal complexes fall into this category.
VerySlowConv Selects appropriate SCF converger cri-
teria for very difficult cases.
SCFConvForced Force convergence: do not continue with
the calculation, if the SCF did not con-
verge.
CPCM(solvent) CPCM Invoke the conductor-like polarizable
C-PCM continuum model with a standard sol-
vent (see section 9.27 for a list of sol-
vents). If no solvent is given, infinity (a
conductor) is assumed.
Spin-orbit coupling
SOMF(1X) REL SOCType, Invokes the RI-SOMF(1X) treatment of
SOCFlags the spin-orbit coupling operator
Miscellaneous options
ANGS COORDS UNITS Select angstrom units
BOHRS Select input coordinates in atomic units
FRACOCC SCF FRACOCC Turns the fractional occupation option
on (FOD is always calculated in this
case)
SMEAR SCF SMEARTEMP Temperature for occupation number
smearing on (default is 5000 K; FOD
(see 9.5.8.2) is always calculated in this
case)
NOSMEAR Turn occupation number smearing off
KEEPINTS SCF KEEPINTS Keep two electron integrals on disk
NOKEEPINTS Do not keep two electron integrals
KEEPDENS SCF KEEPDENS Keep the density matrix on disk
NOKEEPDENS Do not keep the density matrix
READINTS SCF READINTS Reading of two electron integrals on
NOREADINTS Reading of two electron integrals off
CHEAPINTS SCF USECHEAPINTS Use the cheap integral feature in direct
SCF calculations
NOCHEAPINTS Turn that feature off
FLOAT SCF VALFORMAT Set storage format for numbers to single

precision (SCF, RI-MP2, CIS, CIS(D),
MDCI)
DOUBLE SCF VALFORMAT Set storage format for numbers to dou-
ble precision (default)
UCFLOAT SCF VALFORMAT Use float storage in the matrix contain-
COMPRESSION ers without data compression
CFLOAT SCF VALFORMAT Use float storage in the matrix contain-
COMPRESSION ers with data compression
UCDOUBLE SCF VALFORMAT Use double storage in the matrix con-
COMPRESSION tainers without data compression
CDOUBLE SCF VALFORMAT Use double storage in the matrix con-
COMPRESSION tainers with data compression
Output control
NORMALPRINT OUTPUT PRINTLEVEL Selects the normal output
MINIPRINT Selects the minimal output
SMALLPRINT Selects the small output
LARGEPRINT Selects the large output
PRINTMOS OUTPUT Print[p MOS] Prints MO coefficients
NOPRINTMOS OUTPUT Suppress printing of MO coefficients
PRINTBASIS OUTPUT Print[p basis] Print the basis set in input format
PRINTGAP OUTPUT Print[p Prints the HOMO/LUMO gap in each
homolumogap] SCF iteration. This may help to detect
convergence problems
ALLPOP OUTPUT Print[. . . ] Turns on all population analysis
NOPOP Turns off all populaton analysis
MULLIKEN Turns on the Mulliken analysis
NOMULLIKEN Turns off the Mulliken analysis
LOEWDIN Turns on the Loewdin analysis
NOLOEWDIN Turns off the Loewdin analysis
MAYER Turns on the Mayer analysis
NOMAYER Turns off the Mayer analysis
NPA Turns on interface for the NPA analysis
using the GENNBO program
NBO Turns on the interface for the NPA plus
NBO analysis with the GENNBO pro-
gram
NONPA Turns off NPA analysis
NONBO Turns of NBO analysis
REDUCEDPOP Prints Loewdin reduced orb.pop per MO
NOREDUCEDPOP Turns this feature off
UNO SCF UNO Produce UHF natural orbitals
AIM Produce a WFN file
XYZFILE OUTPUT XYZFILE Produce an XYZ coordinate file
PDBFILE PDBFILE Produce a PDB file
Compression and storage. The data compression and storage options deserve some comment: in a number
of modules including RI-MP2, MDCI, CIS, (D) correction to CIS, etc. the program uses so called Matrix
Containers. This means that the data to be processed is stored in terms of matrices in files and is accessed
by a double label. A typical example is the exchange operator Kij with matrix elements K ij (a, b) = (ia|jb).
Here the indices i and j refer to occupied orbitals of the reference state and a and b are empty orbitals of
the reference state. Data of this kind may become quite large (formally N 4 scaling). To store the numbers
in single precision cuts down the memory requirements by a factor of two with (usually very) slight loss
in precision. For larger systems one may also gain advantages by also compressing the data (e.g. use a
packed storage format on disk). This option leads to additional packing/unpacking work and adds some
overhead. For small molecules UCDOUBLE is probably the best option, while for larger molecules UCFLOAT
or particularly CFLOAT may be the best choice. Compression does not necessarily slow the calculation
down for larger systems since the total I/O load may may be substantially reduced and thus (since CPU
is much faster than disk) the work of packing and unpacking takes less time than to read much larger files
(the packing may reduce disk requirements for larger systems by approximately a factor of 4 but it has not
been extensively tested so far). There are many factors contributing to the overall wall clock time in such
cases including the total system load. It may thus require some experimentation to find out with which set of
options the program runs fastest with.
! CAUTION !
It is possible that FLOAT may lead to unacceptable errors. Thus it is

not the recommended option when MP2 or RI-MP2 gradients or relaxed
densities are computed. For this reason the default is DOUBLE.
If you have convinced yourself that FLOAT is OK, it may save you a
factor of two in both storage and CPU.
Global memory use. Some ORCA modules (in particular those that perform some kind of wavefunction
based correlation calculations) require large scratch arrays. Each module has an independent variable
to control the size of these dominant scratch arrays. However, since these modules are never running
simultaneously, we provide a global variable MaxCore that assigns a certain amount of scratch memory to all
of these modules. Thus:
%MaxCore 4000
sets 4000 MB (= 4 GB) as the limit for these scratch arrays. This limit applies per processing core.
Do not be surprised if the program takes more than that this size only refers to the dominant work areas.
Thus, you are well advised to provide a number that is significantly less than your physical memory. Note
also that the memory use of the SCF program cannot be controlled: it dynamically allocates all memory that
it needs and if it runs out of physical memory you are out of luck. This, however, rarely happens unless you
run on a really small memory computer or you are running a gigantic job.
6.2.2 Density Functional Methods
For density functional calculations a number of standard functionals can be selected via the simple input
feature. Since any of these keywords will select a DFT method, the keyword DFT is not needed in the
input. Further functionals are available via the %method block. References are given in section 9.2.2.1.
Table 6.2: Density functionals available in ORCA.

Local and gradient corrected functionals
HFS HartreeFockSlater Exchange only functional
LDA or LSD Local density approximation (defaults to VWN5)
VWN or VWN5 VoskoWilkNusair local density approx. parameter set V
VWN3 VoskoWilkNusair local density approx. parameter set III
PWLDA Perdew-Wang parameterization of LDA
BP86 or BP Becke 88 exchange and Perdew 86 correlation
BLYP Becke 88 exchange and Lee-Yang-Parr correlation
OLYP Handys optimal exchange and Lee-Yang-Parr correlation
GLYP Gills 96 exchange and Lee-Yang-Parr correlation
XLYP The Xu and Goddard exchange and Lee-Yang-Parr correlation
PW91 Perdew-Wang 91 GGA functional
mPWPW Modified PW exchange and PW correlation
mPWLYP Modified PW exchange and LYP correlation
PBE Perdew-Burke-Erzerhoff GGA functional
RPBE Modified PBE
REVPBE Revised PBE
PWP Perdew-Wang 91 exchange and Perdew 86 correlation
Hybrid functionals
B1LYP The one-parameter hybrid functional with Becke 88 exchange
and Lee-Yang-Parr correlation (25% HF exchange)
B3LYP and B3LYP/G The popular B3LYP functional (20% HF exchange) as defined
in the TurboMole program system and the Gaussian program
system, respectively
O3LYP The Handy hybrid functional
X3LYP The Xu and Goddard hybrid functional
B1P The one-parameter hybrid version of BP86
B3P The three-parameter hybrid version of BP86
B3PW The three-parameter hybrid version of PW91
PW1PW One-parameter hybrid version of PW91
mPW1PW One-parameter hybrid version of mPWPW
mPW1LYP One-parameter hybrid version of mPWLYP
PBE0 One-parameter hybrid version of PBE
PW6B95 Hybrid functional by Truhlar
BHANDHLYP Half-and-half hybrid functional by Becke
Meta-GGA and hybrid meta-GGA functionals
TPSS The TPSS meta-GGA functional
TPSSh The hybrid version of TPSS (10% HF exchange)
6.3 Basis Sets 23
TPSS0 A 25% exchange version of TPSSh that yields improved ener-

getics compared to TPSSh but is otherwise not well tested
M06L The Minnesota M06-L meta-GGA functional
M06 The M06 hybrid meta-GGA (27% HF exchange)
M062X The M06-2X version with 54% HF exchange
Range-separated hybrid functionals
wB97 Head-Gordons fully variable DF B97
wB97X Head-Gordons DF B97X with minimal Fock exchange
wB97X-D3 Chais refit incl. D3 correction
CAM-B3LYP Handys fit
LC-BLYP Hiraos original application
Perturbatively corrected double-hybrid functionals
B2PLYP The new mixture of MP2 and DFT from Grimme
RI-B2PLYP B2PLYP with RI applied to the MP2 part
B2PLYP-D B2PLYP with Grimmes empirical dispersion correction from
2006 (D2) [1]
B2PLYP-D3 B2PLYP with Grimmes atom-pairwise dispersion correction
from 2010 [2] and Becke-Johnson damping (D3BJ)
RI-B2PLYP RIJONX The same but with RI also applied in the SCF part
mPW2PLYP mPW exchange instead of B88 (also with RI and RIJONX as
above for B2PLYP). mPW is supposed to improve on weak
interactions.
mPW2PLYP-D mPW2PLYP with Grimmes empirical dispersion correction
from 2006 (D2)
B2GP-PLYP Gershom Martins general purpose reparameterization
B2K-PLYP Gershom Martins kinetic reparameterization
B2T-PLYP Gershom Martins thermochemistry reparameterization
PWPB95 Hybrid functional with MP2 mixture from Grimme
RI-PWPB95 PWPB95 with RI for the MP2 part
Dispersion corrections (see 8.1.4.6 and 9.2.2.10 for details)
D3BJ Atom-pairwise dispersion correction to the DFT energy with
Becke-Johnson damping (recommended)
D3ZERO Atom-pairwise dispersion correction with zero damping
D2 Empirical dispersion correction from 2006 (not recommended)
6.3 Basis Sets
6.3.1 Standard basis set library
There are standard basis sets that can be specified via the simple input feature in the keyword line.
However, any basis set that is not already included in the ORCA library can be provided either directly
in the input or through an external file. See the BASIS input block for a full list of internal basis sets
and various advanced aspects (section 9.3). Effective core potentials and their use are described in section 6.3.3.
Table 6.3: Basis sets available on ORCA.

Pople-style basis sets
3-21G Pople 3-21G (HCs)
STO-3G Minimal basis set(HI)
3-21GSP Buenker 3-21GSP (HAr)
4-22GSP Buenker 4-22GSP (HAr)
6-31G Pople 6-31G and its modifications (HZn)
m6-31G Modified 6-31G for 3d transition metals (ScCu)
6-311G Pople 6-311G and its modifications (HBr)
Polarization functions for the 6-31G basis set:
* or (d) One set of first polarization functions on all atoms except H
** or (d,p) One set of first polarization functions on all atoms
Further combinations: (2d), (2df), (2d,p), (2d,2p), (2df,2p), (2df,2pd)
Polarization functions for the 6-311G basis set:
All of the above plus (3df) and (3df,3pd)
Diffuse functions for the 6-31G and 6-311G basis sets:
+ before G Include diffuse functions on all atoms except H (e.g. 6-31+G)
++ before G Include diffuse functions on all atoms. Works only when H
polarization is already included, e.g. 6-31++G(d,p)
The def2 basis sets of the Karlsruhe group
These basis sets are all-electron for elements HKr, and automatically load Stuttgart-Dresden
effective core potentials for elements RbRn.
def2-SVP Valence double-zeta basis set with new polarization functions.
def2-SV(P) The above with slightly reduced polarization.
def2-TZVP Valence triple-zeta basis set with new polarization functions.
Note that this is quite similar to the older (def) TZVPP for
the main group elements and TZVP for hydrogen.
def2-TZVP(-f) TZVP with f polarization removed from main group elements.
def2-TZVPP TZVPP basis set with new polarization functions.
def2-QZVPP Accurate polarized quadruple-zeta basis.
Older (def ) Ahlrichs basis sets
All-electron basis sets for elements HKr.
SV Valence double-zeta basis set.
SV(P) Valence double-zeta with polarization only on heavy elements.
SVP Polarized valence double-zeta basis set.
TZV Valence triple-zeta basis set.
TZV(P) Valence triple-zeta with polarization on heavy elements.
TZVP Polarized valence triple-zeta basis set.
TZVPP Doubly polarized triple-zeta basis set.
QZVP Polarized valence quadruple-zeta basis set.
QZVPP Doubly polarized quadruple-zeta basis set.
6.3 Basis Sets 25
Note: Past versions of ORCA used to load all-electron basis sets also for elements RbI with
the above keywords for double- and triple-zeta basis sets. The RbI basis sets originated from
non-relativistic all-electron basis sets of the Turbomole library (such as TZVPAlls). This
automatic substitution is now deprecated. However, we offer temporarily the ability to
reproduce that behavior by adding the prefix old- to the above keywords, e.g. old-TZVP.
Diffuse def2 basis sets
Minimally augmented def2 ba- Augmented def2 basis sets by diffuse s and p functions according
sis sets to Truhlar [3]. Recommended for general use.
ma-def2-SVP Minimally augmented def2-SVP basis set.

ma-def2-SV(P) Minimally augmented def2-SV(P) basis set.
ma-def2-TZVP Minimally augmented def2-TZVP basis set.
ma-def2-TZVP(-f) Minimally augmented def2-TZVP(-f) basis set.
ma-def2-TZVPP Minimally augmented def2-TZVPP basis set.
ma-def2-QZVPP Minimally augmented def2-QZVPP basis set.
Rappoport property-optimized Augmented def2 basis sets by diffuse functions according to

diffuse def2 basis sets Rappoport et al. [4]
def2-SVPD Diffuse def2-SVP basis set for property calculations

def2-TZVPD Diffuse def2-TZVP basis set for property calculations
def2-TZVPPD Diffuse def2-TZVPP basis set for property calculations
def2-QZVPD Diffuse def2-QZVP basis set for property calculations
def2-QZVPPD Diffuse def2-QZVPP basis set for property calculations
Relativistically recontracted Karlsruhe basis sets
For use in DKH or ZORA calculations we provide adapted versions of the def2 basis sets
for the elements HKr (i.e., for the all-electron def2 basis sets). These basis sets retain the
original def2 exponents but have only one contracted function per angular momentum (and
hence are somewhat larger), with contraction coefficients suitable for the respective scalar
relativistic Hamiltonian. These basis sets can be called with the prefix DKH- or ZORA-, and
can be combined with the SARC basis sets for the heavier elements.
DKH-def2-SVP and ZORA-def2-SVP

DKH-def2-SV(P) and ZORA-def2-SV(P)
DKH-def2-TZVP and ZORA-def2-TZVP
DKH-def2-TZVP(-f) and ZORA-def2-TZVP(-f)
DKH-def2-TZVPP and ZORA-def2-TZVPP
DKH-def2-QZVPP and ZORA-def2-QZVPP
Minimally augmented versions:
ma-DKH-def2-SVP and ma-ZORA-def2-SVP
ma-DKH-def2-SV(P) and ma-ZORA-def2-SV(P)
ma-DKH-def2-TZVP and ma-ZORA-def2-TZVP
ma-DKH-def2-TZVP(-f) and ma-ZORA-def2-TZVP(-f)
ma-DKH-def2-TZVPP and ma-ZORA-def2-TZVPP
ma-DKH-def2-QZVPP and ma-ZORA-def2-QZVPP
The same functionality is offered for the def basis sets, e.g. ZORA-TZVP. In this case
too, the relativistically recontracted versions refer to the elements HKr. To replicate the
behavior of past ORCA versions for elements RbI, the prefix old- can be used with these
keywords as in the non-relativistic case.
WARNING: Previous verions of ORCA made extensive use of automatic basis

set substitution and aliasing when the use of the DKH or ZORA Hamiltonians
was detected. This is no longer the case! Relativistic versions of Karlsruhe basis
sets now have to be requested explicitly with the appropriate prefix. SARC basis
sets also have to be requested explicitly.
SARC basis sets [59]
Segmented all-electron relativistically contracted basis sets for use with the DKH2 and ZORA
Hamiltonians. Available for elements beyond Xe.
SARC-DKH-TZVP
SARC-DKH-TZVPP
SARC-ZORA-TZVP
SARC-ZORA-TZVPP
SARC-DKH-SVP and SARC-ZORA-SVP are also available for the 5d transition metals only.
Note: SARC/J is the general-purpose Coulomb-fitting auxiliary for all SARC orbital basis
sets.
SARC2 basis sets for the lanthanides [10]
SARC basis sets of quadruple-zeta quality for lanthanides. With polarization they are suitable
for accurate calculations using correlated wave function methods.
SARC2-DKH-QZV SARC2 basis set of valence quadruple-zeta quality.
SARC2-DKH-QZVP Extended with NEVPT2-optimized (3g2h) polarization.
SARC2-ZORA-QZV
SARC2-ZORA-QZVP
Note: Each of the above basis sets has a large dedicated /JK auxiliary basis set for simulta-
neous Coulomb and exchange fitting.
Jensen basis bets
pc-n (n = 0, 1, 2, 3, 4) Polarization-consistent generally contracted
basis sets (HKr) of up to quintuple-zeta quality, optimized for
SCF calculations
aug-pc-n As above, augmented by diffuse functions
pcseg-n Segmented PC basis sets (HKr), DFT-optimized
aug-pcseg-n As above, augmented by diffuse functions
pcSseg-n Segmented contracted basis sets (HKr) optimized for nuclear
magnetic shielding
aug-pcSseg-n As above, augmented by diffuse functions
pcJ-n Segmented contracted basis sets (HAr) optimized for spin-spin
coupling constants
aug-pcJ-n As above, augmented by diffuse functions
6.3 Basis Sets 27
Sapporo basis bets

Sapporo-nZP-2012 (n = D, T, Q) All-electron generally contracted non-relativistic
basis sets (HXe)
Sapporo-DKH3-nZP-2012 (n = D, T, Q) All-electron basis sets optimized for the DKH3
Hamiltonian and finite nucleus (KRn)
Correlation-consistent basis sets
cc-pVDZ Dunning correlation-consistent polarized double-zeta
cc-pVTZ Dunning correlation-consistent polarized triple-zeta
cc-pVQZ Dunning correlation-consistent polarized quadruple-zeta
cc-pV5Z Dunning correlation-consistent polarized quintuple-zeta
cc-pV6Z Dunning correlation-consistent polarized sextuple-zeta
aug-cc-pVnZ (n = D, T, Q, 5, 6) Augmented with diffuse functions
cc-pCVnZ (n = D, T, Q, 5, 6) Core-polarized basis sets
aug-cc-pCVnZ (n = D, T, Q, 5, 6) as above, augmented with diffuse functions
cc-pwCVnZ (n = D, T, Q, 5) Core-polarized with weighted core functions
aug-cc-pwCVnZ (n = D, T, Q, 5) as above, augmented with diffuse functions
cc-pVn(+d)Z (n = D, T, Q, 5) with tight d functions
DKH versions of correlation-consistent basis sets
cc-pVnZ-DK (n = D, T, Q, 5) Correlation-consistent all-electron basis sets
for use with the 2nd-order Douglas-Kroll-Hess Hamiltonian
aug-cc-pVnZ-DK (n = D, T, Q, 5) as above, augmented with diffuse functions
cc-pwCVnZ-DK (n = D, T, Q, 5) DK versions of weighted core correlation-
consistent basis sets
aug-cc-pwCVnZ-DK (n = D, T, Q, 5) weighted-core DK basis sets with diffuse
functions
ECP-based versions of correlation-consistent basis sets
cc-pVnZ-PP (n = D, T, Q, 5) Correlation-consistent all-electron basis sets
combined with SK-MCDHF-RSC effective core potentials
aug-cc-pVnZ-PP (n = D, T, Q, 5) as above, augmented with diffuse functions
cc-pwCVnZ-PP (n = D, T, Q, 5) with weighted core functions
aug-cc-pwCVnZ-PP (n = D, T, Q, 5) as above, augmented with diffuse functions
F12 and F12-CABS basis sets
cc-pVnZ-F12 (n = D, T, Q) Special orbital basis sets for F12 calculations
(larger than the regular D, T, Q-zeta basis sets!)
cc-pCVnZ-F12 (n = D, T, Q) with core polarization functions
cc-pVnZ-PP-F12 (n = D, T, Q) ECP-based versions
cc-pVnZ-F12-CABS (n = D, T, Q) Near-complete auxiliary basis sets for F12
calculations
cc-pVnZ-F12-OptRI (n = D, T, Q)
cc-pCVnZ-F12-OptRI (n = D, T, Q)
cc-pVnZ-PP-F12-OptRI (n = D, T, Q)
aug-cc-pVnZ-PP-F12-OptRI (n = D, T, Q, 5)
aug-cc-pwCVnZ-PP-F12- (n = D, T, Q, 5)
OptRI
Atomic Natural Orbital basis sets
ano-pVnZ (n = D, T, Q, 5). Our newly contracted ANO basis sets on
the basis of the cc-pV6Z (or pc-4 where missing) primitives.
These are very accurate basis sets that are significantly better
than the cc-pVnZ counterparts for the same number of basis
functions (but much larger number of primitives of course).
saug-ano-pVnZ (n = D, T, Q) augmentation with a single set of sp functions.
Greatly enhances the accuracy of the SCF energies but not for
correlation energies.
aug-ano-pVnZ (n = D, T, Q) full augmentation with spd, spdf, spdfg set of
polarization functions. Almost as expensive as the next higher
basis set. In fact, aug-ano-pVnZ = ano-pV(n + 1)Z with the
highest angular momentum polarization function deleted.
Relativistic contracted ANO-RCC basis sets
ANO-RCC-FULL The complete ANO-RCC basis sets (H-Cm). Some default
contractions are provided for convenience with the keywords:
ANO-RCC-DZP
ANO-RCC-TZP
ANO-RCC-QZP
Miscellaneous and specialized basis sets
D95 Dunnings double-zeta basis set (HCl).
D95p Polarized version of D95.
MINI Huzinagas minimal basis set.
MINIS Scaled version of the MINI.
MIDI Huzinagas valence double-zeta basis set.
MINIX Combination of small basis sets by Grimme (see Table 9.5).
Wachters+f First-row transition metal basis set (ScCu).
Partridge-n (n = 1, 2, 3, 4) Uncontracted basis sets by Partridge.
LANL2DZ Los Alamos valence double-zeta with HayWadt ECPs.

LANL2TZ Triple-zeta version.
LANL2TZ(f) Triple-zeta plus polarization.
LANL08 Uncontracted basis set.
LANL08(f) Uncontracted basis set + polarization.
EPR-II Barones basis set (H, BF) for EPR calculations (double-zeta).
EPR-III Barones basis set for EPR calculations (triple-zeta).
IGLO-II Kutzelniggs basis set (H, BF, AlCl) for NMR and EPR
calculations.
IGLO-III Larger version of the above.
aug-cc-pVTZ-J Sauers basis set for accurate hyperfine coupling constants.
Auxiliary basis sets. Auxiliary basis sets for the RI-J and RI-MP2 approximations can also be specified
directly in the simple input:
6.3 Basis Sets 29
Table 6.4: Overview of auxiliary basis sets available in ORCA.

Auxiliary basis sets for Coulomb fitting
Def2/J Weigends universal Coulomb fitting basis that is suitable for
all def2 type basis sets. Assumes the use of ECPs beyond Kr (do
not use with DKH/ZORA).
SARC/J General-purpose Coulomb fitting basis set for all-electron calcula-
tions. Consists of the decontracted def2/J up to Kr and of our own
auxiliary basis sets for the rest of the periodic table. Appropriate
for use in DKH or ZORA calculations with the recontracted ver-
sions of the all-electron def2 basis sets (up to Kr) and the SARC
basis sets for the heavier elements.
Auxiliary basis sets for simultaneously fitting Coulomb and exchange
Fitting basis sets developed by Weigend for fitting simultaneously Coulomb and exchange energies.
They are quite large and accurate. They fit SCF energies very well but even if they are large they
do not fit correlation as well as the dedicated /C auxiliary basis sets.
Def2/JK Coulomb+Exchange fitting for all def2 basis sets
Def2/JKsmall reduced version of the above
cc-pVnZ/JK (n = T, Q, 5) for the respective cc-pVnZ orbital basis
aug-cc-pVnZ/JK (n = T, Q, 5) for the respective aug-cc-pVnZ orbital basis
Auxiliary basis sets for correlation calculations
Def2-SVP/C Correlation fitting for the def2-SVP orbital basis
Def2-TZVP/C for the def2-TZVP orbital basis
Def2-TZVPP/C for the def2-TZVPP orbital basis
Def2-QZVPP/C for the def2-QZVPP orbital basis
cc-pVnZ/C (n = D, T, Q, 5, 6) for the respective cc-pVnZ orbital basis
aug-cc-pVnZ/C (n = D, T, Q, 5, 6) for the respective aug-cc-pVnZ orbital basis
cc-pwCVnZ/C (n = D, T, Q, 5) for the respective cc-pwCVnZ orbital basis
aug-cc-pwCVnZ/C (n = D, T, Q, 5) for the respective aug-cc-pwCVnZ orbital basis
cc-pVnZ-PP/C (n = D, T, Q) for the respective cc-pVnZ-PP orbital basis
aug-cc-pVnZ-PP/C (n = D, T, Q) for the respective aug-cc-pVnZ-PP orbital basis
cc-pwCVnZ-PP/C (n = D, T, Q) for the respective cc-pwCVnZ-PP orbital basis
aug-cc-pwCVnZ-PP/C (n = D, T, Q) for the respective aug-cc-pwCVnZ-PP orbital basis
cc-pVnZ-F12-MP2fit (n = D, T, Q) for the respective cc-pVnZ-F12 orbital basis
cc-pCVnZ-F12-MP2fit (n = D, T, Q) for the respective cc-pCVnZ-F12 orbital basis
cc-pVnZ-PP-F12-MP2fit (n = D, T, Q) for the respective cc-pVnZ-PP-F12 orbital basis
AutoAux Automatic construction of a general purpose auxiliary basis for
simultaneously fitting Coulomb, exchange and correlation calcula-
tions. See section 9.3.1 for details.
NOTE: ORCA versions before 4.0 allowed the use of multiple keywords to invoke the same def2 Coulomb or
Coulomb+exchange fitting basis set of Weigend. To avoid confusion all these keywords are now deprecated
and the auxiliary basis sets are simply called using def2/J and def2/JK.
NOTE: Starting from version 2.6.63 ORCA can deal with two auxiliary basis sets one for Coulomb and
one for correlation. The default is the Coulomb fitting aux-basis. If you select a separate correlation fitting
basis, the correlation modules (RI-MP2, RI-MDCI, RI-MRCI) will replace this Coulomb fitting aux-basis
with the correlation fitting basis. In order to use this feature you have to give the /C-fit basis via the
simple input lines.
6.3.2 Use of scalar relativistic basis sets
For DKH and ZORA calculations ORCA provides relativistically recontracted versions of the Karlsruhe basis
sets for elements up to Kr. These can be requested by adding the prefix DKH- or ZORA- to the normal basis
set name. Note that for other non-relativistic basis sets (for example Pople-style bases) no recontraction has
been performed and consequently such calculations are inconsistent! The basis set and the scalar relativistic
Hamiltonian are specified in the keyword line, for example:
! B3LYP ZORA ZORA-TZVP ...
If an auxiliary basis set is required for these recontracted Karlsruhe basis sets, we recommend the use of
the decontracted def2/J. This can be obtained simply by using the keyword ! SARC/J (instead of the
equivalent ! def2/J DecontractAux) and is the recommended option as it simultaneously covers the use
of SARC basis sets for elements beyond Krypton.
! TPSS ZORA ZORA-def2-TZVP SARC/J ...
For all-electron calculations with heavier elements (third-row transition metals, lanthanides, actinides and
6p elements) we offer the SARC (segmented all-electron relativistically contracted) basis sets [59]. These
were specifically developed for scalar relativistic calculations and are individually adapted to the DKH2 and
ZORA Hamiltonians. In this case the auxiliary basis set must be specified as SARC/J.
! PBE DKH SARC-DKH-TZVP SARC/J ...
Other basis sets suitable for scalar relativistic calculations are various versions of the all-electron correlation-
consistent basis sets that are optimized for the DKH2 Hamiltonian and can be called with the suffix -DK.
The relativistically contracted atomic natural orbital (ANO-RCC) basis sets of Roos and coworkers were also
developed for the DKH2 Hamiltonian and have almost complete coverage of the periodic table (up to Cm).
6.3.3 Effective Core Potentials
Starting from version 2.8.0, ORCA features effective core potentials (ECPs). They are a good alternative
to scalar relativistic all-electron calculations if heavy elements are involved. This pertains to geometry
optimizations and energy calculations but may not be true for property calculations.
In order to reduce the computational effort, the usually highly contracted and chemically inert core basis
functions can be eliminated by employing ECPs. ECP calculations comprise a valence-only basis and thus
are subject to the frozen core approximation. Contributions due to the core orbitals are accounted for by an
effective one-electron operator U core which replaces the interactions between core and valence electrons and
6.3 Basis Sets 31
accounts for the indistinguishability of the electrons. Its radial parts Ul (r) are generally expressed as a linear
combination of Gaussian functions, while the angular dependence is included through angular momentum
l
projectors |Sm i.
L1
X l
X
U core = UL (r) +
l l

Sm i [Ul (r) UL (r)] hSm
l=0 m=l
X
Ul = dkl rnkl exp(kl r2 )
k
atom
The maximum angular momentum L is generally defined as lmax + 1. The parameters nkl , kl and dkl
that are necessary to evaluate the ECP integrals have been published by various authors, among them the
well-known Los Alamos (LANL) [11] and StuttgartDresden (SD) [1257] parameter sets. Depending on
the specific parametrization of the ECP, relativistic effects can be included in a semiempirical fashion in an
otherwise nonrelativistic calculation. Introducing U core into the electronic Hamiltonian yields two types of
ECP integrals, the local (or type-1) integrals that arise because of the maximum angular momentum potential
UL and the semi-local (or type-2) integrals that result from the projected potential terms. The evaluation of
these integrals in ORCA proceeds according to the scheme published by Flores-Moreno et al. [58].
A selection of ECP parameters and associated basis sets is directly accessible in ORCA through the internal
ECP library (see table 6.5 for a listing of keywords).
Table 6.5: Overview of library keywords for ECPs and associated basis sets available in ORCA.
ECP Appendant Elements Valence basis sets
keyword ECP1
StuttgartDresden ECPs

available for RbLa, HfRn:

def2-SD SD(28,MWB) RbCd

SD(28,MDF) InXe2 def2-SV(P), def2-SVP, def2-TZVP,

SD(46,MWB) CsLa def2-TZVPP, def2-QZVP, def2-QZVPP

SD(60,MWB) HfPt

SD(60,MDF) AuRn3

auxiliary basis sets:

def2/J, def2/JK, def2-SVP/C, etc.

def-SD SD(28,MWB) RbCd

SD(46,MWB) InLa available for RbLr:

SD(28,MWB) CeLu

SD(60,MWB) HfPt def-SV(P), def-SVP, def-TZVP,
SD(60,MDF) Au,3 Hg,3 Rn

def-TZVPP, def-QZVP, def-QZVPP

SD(78,MWB) TlAt

SD(78,MDF) Fr, Ra auxiliary basis sets:

SD(60,MWB) AcLr def2/J, def2/JK, def2-SVP/C, etc.

SDD SD(2,SDF) Li, Be SD(2,SDF)

SD(2,MWB) BNe SD(2,MWB)
SD(10,SDF) Na, Mg SD(10,SDF)
SD(10,MWB) AlCa SD(10,MWB)
SD(10,MDF) ScZn RSC4
SD(28,MWB) GaCd GaSr: SD(28,MWB), YCd: SD(28,MHF)
SD(28,MDF) GeBr, RbXe SD(28,MDF)

SD(46,MWB) InBa SD(46,MWB)
SD(28,MWB) LaLu SD(28,MWB)-SEG
SD(60,MWB) HfHg RSC5
SD(78,MWB) TlRn SD(78,MWB)
SD(60,MWB) AcLr RSC6
Los Alamos ECPs
LANL2 LANL(10) KCu LANL-type basis sets:
LANL(28) RbAg LANL2DZ, LANL2TZ, LANL08
LANL(46) CsLa
LANL(60) HfAu
LANL1 LANL(10) NaAr
LANL(18) KZn
LANL(28) GaKr
LANL(36) RbCd
LANL(46) InXe
LANL(54) CsLa
LANL(68) HfTl
LANL(78) Pb, Bi
DEF2-ECP ecp-28 YCd
ecp-46 HfHg
ecp-28 InSb
ecp-46 TlBi
ecp-28 Te-Xe
ecp-46 Po-Rn
ecp-28 Rb
ecp-46 Cs
ecp-28 Sr
ecp-46 Ba
HAYWADT- Na Bi
ECP
SK- Cu - Rn
MCDHF-
RSC-ECP
1
The ECP names in the second column of the table consist of a token (SD, LANL), followed by the
number of replaced core electrons in parentheses and, if necessary, tags for the reference method
and data (S: single-valence-electron ion; M: neutral atom; HF: HartreeFock; WB: quasi-relativistic;
DF: relativistic).
2
I: OLD-SD(28,MDF) due to compatibility with TURBOMOLE.
3
Au, Hg: OLD-SD(60,MDF) due to compatibility with TURBOMOLE.
4
ScNi: truncated and modified SD(10,MDF) basis set. Cu, Zn: truncated SD(10,MWB) basis set.
5
Truncated SD(60,MWB) basis set.
6
Consider using SD(60,MHF), which has the same contraction scheme.
By default, the Def2-X basis sets use the Def2-ECP effective core potential definition!
The same holds true for the auxiliary basis sets used in the same context!
6.4 ORCA and Symmetry 33
The simplest way to assign ECPs and associated valence basis sets is by using the ECP keyword within the
keyword line. The ECP keyword itself assigns only the effective core potential, not a valence basis set!
As an example for an explicitly named ECP you could use
! def2-TZVP def2-sd
This would assign the def2-sd ecp according to the definition given in the table above. Without the def2-sd
keyword ORCA would default to Def2-ECP.
The old behaviour, that is, pre ORCA 4.0, in which a valence basis set gets automatically assigned, is still
available for now.
%basis oldecp "SDD"

end
In this case, all elements will receive the ECP as shown in the table above, as well as the valence basis set
shown!
6.4 ORCA and Symmetry
For most of its life, ORCA did not take advantage of molecular symmetry. Starting from version 2.8.0 there
is at least limited use. On request, with the UseSym keyword, the program detects the point group, cleans up
the coordinates, orients the molecule and produces symmetry-adapted orbitals in SCF/CASSCF calculations.
Note however that the calculation time will not be reduced. Only D2h and subgroups are currently supported.
The only correlation module that makes use of this information so far is the MRCI module. Here and in
CASSCF calculations, the use of symmetry helps to control the calculation and the interpretation of the
results. More symmetry is likely to be implemented in the future, although it is unlikely that the program
will ever take advantage of symmetry in a very big way.
If the automatic symmetry detection fails to find the expected point group, the coordinates specified are not
absolutely symmetrical to that group, and one should take a careful look at the input coordinates, maybe
using a visualization program. A problem often encountered when using coordinates generated from other
jobs (e.g. geometry optimizations) is the detection of a too low symmetry because of numerical noise. This
can be solved by increasing the detection threshold using an input line which looks like this:
%method SymThresh 5.0e-2 end
However, it is not recommended to run calculations on a very high threshold, since this may introduce some
odd behavior. Instead, a method to symmetrize the coordinates is to do a fake run with NoIter, XYZFile
and an increased threshold, and then to use the created .xyz file as input for the actual calculation. This
has the additional benefit that the input coordinates stored in your data are already symmetrical. To give
an example: the following coordinates for staggered ethane were obtained by geometry optimization NOT
using the symmetry module. They are, however, not recognized as D3d symmetrical due to numerical noise
and instead are found to be of Ci symmetry (a subgroup of D3d ). To counter this, the detection threshold is
increased and a symmetry perfected coordinate file is produced by the following input:
! RHF SVP UseSym NoIter XYZfile

%method SymThresh 1.0e-2 end
*xyz 0 1
C -0.002822 -0.005082 -0.001782
C -0.723141 -1.252323 -0.511551
H 0.017157 0.029421 1.100049
H 1.042121 0.030085 -0.350586
H -0.495109 0.917401 -0.350838
H -0.743120 -1.286826 -1.613382
H -0.230855 -2.174806 -0.162495
H -1.768085 -1.287489 -0.162747
*
6.5 Jobs with Multiple Steps
ORCA supports input files with multiple jobs. This feature is designed to simplify series of closely related
calculations on the same molecule or calculations on different molecules. The objectives for implementing
this feature include:
Calculate of a molecular property using different theoretical methods and/or basis sets for one molecule.
Calculations on a series of molecules with identical settings.
Geometry optimization followed by more accurate single points and perhaps property calculations.
Crude calculations to provide good starting orbitals that may then be used for subsequent calculations
with larger basis sets.
For example consider the following job that in the first step computes the g-tensor of BO at the LDA level,
and in the second step using the BP86 functional.
# -----------------------------------------------------
! UKS LSD SVP grid5 NoFinalGrid TightSCF KeepInts
# -----------------------------------------------------
%eprnmr gtensor 1 end
* int 0 2
B 0 0 0 0 0 0
O 1 0 0 1.2049 0 0
*
6.5 Jobs with Multiple Steps 35
# *************************************************
# ****** This starts the input for the next job *
# *************************************************
$new_job
# --------------------------------------------------
! BP86 SVP SmallPrint ReadInts NoKeepInts
# --------------------------------------------------
%eprnmr gtensor 1 end
* int 0 2
B 0 0 0 0 0 0
O 1 0 0 1.2049 0 0
*
What happens if you use the $new job feature is that all calculation flags for the actual job are transferred
from the previous job and that only the changes in the settings must be input by the user. Thus if you turn
on some flags for one calculation that you do not want for the next, you have to turn them off again yourself
(for example the use of the RI approximation)! In addition, the default is that the new job takes the orbitals
from the old job as input. If you do not want this you have to overwrite this default by specifying your
desired guess explicitly.
Changing the default BaseName
Normally the output files for MyJob.inp are returned in MyJob.xxx (any xxx, for example xxx=out).
Sometimes, and in particular in multistep jobs, you will want to change this behavior. To this end there is
the variable %base that can be user controlled. All filenames (also scratch files) will then be based on this
default name.
36
7 Input of Coordinates
Coordinates can be either specified directly in the input file or read from an external file, and they can be in
either Cartesian (xyz) or internal coordinate format (Z-matrix).
7.1 Reading coordinates from the input file
The easiest way to specify coordinates in the input file is by including a block like the following, enclosed by
star symbols:
* CType Charge Multiplicity

...
coordinate specifications
...
*
Here CType can be one of xyz, int (or internal), or gzmt, which correspond to Cartesian coordinates,
internal coordinates, and internal coordinates in Gaussian Z-matrix format.
The input of Cartesian coordinates in the xyz option is straightforward. Each line consists of the label
for a given atom type and three numbers that specify the coordinates of the atom. The units can be either
Angstrom or Bohr. The default is to specify the coordinates in

Angstr
oms (this can be changed through the
keyword line or via the variable Units in the %coords main block described below).
* xyz Charge Multiplicity

Atom1 x1 y1 z1
Atom2 x2 y2 z2
...
*
For example for CO+ in a S = 1/2 state (multiplicity = 2 1/2 + 1 = 2)
* xyz 1 2
C 0.0 0.0 0.0
O 0.0 0.0 1.1105
*
7.1 Reading coordinates from the input file 37
Internal coordinates are specified in the form of the familiar Z-matrix. A Z-matrix basically contains
information about molecular connectivity, bond lengths, bond angles and dihedral angles. The program then
constructs Cartesian coordinates from this information. Both sets of coordinates are printed in the output
such that conversion between formats is facilitated. The input in that case looks like:
* int Charge Multiplicity

Atom1 0 0 0 0.0 0.0 0.0
Atom2 1 0 0 R1 0.0 0.0
Atom3 1 2 0 R2 A1 0.0
Atom4 1 2 3 R3 A2 D1
. . .
AtomN NA NB NC RN AN DN
*
The rules for connectivity in the internal mode are as follows:
NA: The atom that the actual atom has a distance (RN) with.
NB: The actual atom has an angle (AN) with atoms NA and NB.
NC: The actual atom has a dihedral angle (DN) with atoms NA, NB and NC. This is the angle between
the actual atom and atom NC when looking down the NA-NB axis.
Angles are always given in degrees! The rules are compatible with those used in the well known MOPAC and
ADF programs.
Finally, gzmt specifies internal coordinates in the format used by the Gaussian program. This resembles the
following:
* gzmt 0 1
C
O 1 4.454280
Si 2 1.612138 1 56.446186
O 3 1.652560 2 114.631525 1 -73.696925
C 4 1.367361 3 123.895399 2 -110.635060
...
*
An alternative way to specify coordinates in the input file is through the use of the %coords block, which is
organized as follows:
%coords
CTyp xyz # the type of coordinates = xyz or internal
Charge 0 # the total charge of the molecule
Mult 2 # the multiplicity = 2S+1
38 7 Input of Coordinates
Units Angs # the unit of length = angs or bohrs
# the subblock coords is for the actual coordinates

# for CTyp=xyz
coords
Atom1 x1 y1 z1
Atom2 x2 y2 z2
end
# for CTyp=internal
coords
Atom1 0 0 0 0.0 0.0 0.0
Atom2 1 0 0 R1 0.0 0.0
Atom3 1 2 0 R2 A1 0.0
Atom4 1 2 3 R3 A2 D1
. . .
AtomN NA NB NC RN AN DN
end
end
7.2 Reading coordinates from external files
It is also possible to read the coordinates from external files. The most common format is a .xyz file, which
can in principle contain more than one structure (see section 8.2.9 for this multiple XYZ feature):
* xyzfile 1 2 mycoords.xyz
A lot of graphical tools like Gabedit, molden or Jmol can write Gaussian Z-Matrices (.gzmt). ORCA can
also read them from an external file with the following
* gzmtfile 1 2 mycoords.gzmt
Note that if multiple jobs are specified in the same input file then new jobs can read the coordinates from
previous jobs. If no filename is given as fourth argument then the name of the actual job is automatically
used.
... specification for the first job
$new_job
! keywords
* xyzfile 1 2
In this way, optimization and single point jobs can be very conveniently combined in a single, simple input
file. Examples are provided in the following sections.
7.3 Special definitions 39
7.3 Special definitions
Dummy atoms are defined in exactly the same way as any other atom, by using DA as the atomic
symbol.
Ghost atoms are specified by adding : right after the symbol of the element (see 8.1.6).
Point charges are specified with the symbol Q, followed by the charge (see 9.1.3).
Embedding potentials are specified by adding a > right after the symbol of the element (see
9.3.7).
Non-standard isotopes or nuclear charges are specified with the statements M = . . . and Z =
. . . , respectively, after the atomic coordinate definition.
Fragments can be conveniently defined by declaring the fragment number a given atom belongs to in
parentheses (n) following the element symbol (see 9.1.1).
40
8 Running Typical Calculations
Before entering the detailed documentation of the various features of ORCA it is instructive to provide a
chapter that shows how typical tasks may be performed. This should make it easier for the user to get
started on the program and not get lost in the details of how-to-do-this or how-to-do-that. We hope that the
examples are reasonably intuitive.
8.1 Single Point Energies and Gradients
8.1.1 Hartree-Fock
8.1.1.1 Standard Single Points
In general single point calculations are fairly easy to run. What is required is the input of a method, a basis
set and a geometry. For example, in order run a single point Hartree-Fock calculation on the CO molecule
with the SVP basis set type:
#
# My first ORCA calculation :-)
#
! HF SVP
* xyz 0 1
C 0 0 0
O 0 0 1.13
*
As an example consider this simple calculation on the cyclohexane molecule that may serve as a prototype
for this type of calculation.
# Test a simple direct HF calculation

! HF RHF SV(P)
* xyz 0 1
C -0.79263 0.55338 -1.58694
C 0.68078 0.13314 -1.72622
C 1.50034 0.61020 -0.52199
8.1 Single Point Energies and Gradients 41
C 1.01517 -0.06749 0.77103

C -0.49095 -0.38008 0.74228
C -1.24341 0.64080 -0.11866
H 1.10490 0.53546 -2.67754
H 0.76075 -0.97866 -1.78666
H -0.95741 1.54560 -2.07170
H -1.42795 -0.17916 -2.14055
H -2.34640 0.48232 -0.04725
H -1.04144 1.66089 0.28731
H -0.66608 -1.39636 0.31480
H -0.89815 -0.39708 1.78184
H 1.25353 0.59796 1.63523
H 1.57519 -1.01856 0.93954
H 2.58691 0.40499 -0.67666
H 1.39420 1.71843 -0.44053
*
8.1.1.2 Basis Set Options
There is extensive flexibility in the specification of basis sets in ORCA. First of all, you are not only restricted
to the basis sets that are built in ORCA, but can also read basis set definitions from files. In addition there
is a convenient way to change basis sets on certain types of atoms or on individual atoms. Consider the
following example:
# CuCl4
! UHF HF
%basis basis "SV"
newGTO Cl "DUNNING-DZP" end
end
* xyz -2 2
Cu 0 0 0 newGTO "TZVPP" end
Cl 2.25 0 0
Cl -2.25 0 0
Cl 0 2.25 0
Cl 0 -2.25 0
*
In this example the basis set is initialized as the Ahlrichs split valence basis. Then the basis set on all
atoms of type Cl is changed to DUNNING-DZP and finally the basis set for only the copper atom is changed
to the more accurate TZVPP set. In this way you could treat different atom types or even individual groups
in a molecule according to the desired accuracy. Similar functionality regarding per-element or per-atom
assignments exists for effective core potentials. More details are provided in section 9.3.
42 8 Running Typical Calculations
Sometimes you will like to change the ordering of the starting orbitals to obtain a different electronic state in
the SCF calculation. For example, if we take the last input and want to converge to a ligand field excited
state this can be achieved by:
! UHF HF SV
%basis newGTO Cl "Dunning-DZP" end
end
%scf rotate {48, 49, 90, 1, 1} end
end
* xyz -2 2
Cu 0 0 0 newGTO "TZVPP" end
Cl 2.25 0 0
Cl -2.25 0 0
Cl 0 2.25 0
Cl 0 -2.25 0
*
In the present case, MO 48 is the spin-down HOMO and MO49 the spin-down LUMO. Since we do a
calculation on a Cu(II) complex (d9 electron configuration) the beta LUMO corresponds with the SOMO.
Thus, by changing the SOMO we proceed to a different electronic state (in this case the one with the hole
in the dxy orbital instead of the dx2 y2 orbital). The interchange of the initial guess MOs is achieved by
the command rotate {48, 49, 90, 1, 1} end. What this does is the following: take the initial guess MOs
48 and 49 and rotate them by an angle of 90 degree (this just interchanges them). The two last numbers
mean that both orbitals are from the spin-down set. For RHF or ROHF calculations the operator would be 0.
In general you would probably first take a look at the initial guess orbitals before changing them.
8.1.1.3 SCF and Symmetry
Upon request, the SCF program produces symmetry adapted orbitals. This can help to converge the SCF
on specific excited states of a given symmetry. Take for example the cation H2 O+ : We first run the simple
job:
! SVP UseSym
* xyz 1 2
O 0.000000 0.000000 0.068897
H 0.000000 0.788011 -0.546765
H 0.000000 -0.788011 -0.546765
*
The program will recognize the C2v symmetry and adapt the orbitals to this:
------------------
SYMMETRY DETECTION
------------------
Preparing Data ... done
Detection Threshold: SymThresh ... 1.0000e-04
Point Group will now be determined:

Moving molecule to center of mass ... done
POINT GROUP ... C2v
The coordinates will now be cleaned:

Moving to standard coord frame ... done
(Changed main axis to z and one of the symmetry planes to xz plane)
Structure cleanup requested ... yes
Selected point group ... C2v
Cleaning Tolerance SymThresh ... 1.0000e-04
Some missing point group data is constructed:

Constructing symmetry operations ... done
Creating atom transfer table ... done
Creating asymmetric unit ... done
Cleaning coordinates ... done
-----------------------------------------------
SYMMETRY-PERFECTED CARTESIAN COORDINATES (A.U.)
-----------------------------------------------
0 O 0.00000000 0.00000000 0.13019595
1 H 0.00000000 1.48912498 -1.03323662
2 H 0.00000000 -1.48912498 -1.03323662
------------------
SYMMETRY REDUCTION
------------------
ORCA supports only abelian point groups.
It is now checked, if the determined point group is supported:
Point Group ( C2v ) is ... supported
(Re)building abelian point group:

Creating Character Table ... done
Making direct product table ... done
----------------------
ASYMMETRIC UNIT IN C2v
----------------------
# AT MASS COORDS BAS
0 O 15.9990 0.00000000 0.00000000 0.13019595 0
1 H 1.0080 0.00000000 1.48912498 -1.03323662 0
----------------------
SYMMETRY ADOPTED BASIS
----------------------
The coefficients for the symmetry adopted linear combinations (SALCS)
of basis functions will now be computed:
Number of basis functions ... 24
Preparing memory ... done

Constructing Gamma(red) ... done
Reducing Gamma(red) ... done
Constructing SALCs ... done
Checking SALC integrity ... nothing suspicious
Normalizing SALCs ... done
Storing the symmetry object:

Symmetry file ... Test-SYM-H2O+.sym.tmp
Writing symmetry information ... done
The initial guess in the SCF program will then recognize and freeze the occupation numbers in each irreducible
representation of the C2v point group.
The symmetry of the initial guess is 2-B1

Irrep occupations for operator 0
A1 - 3
A2 - 0
B1 - 1
B2 - 1
A1 - 3
A2 - 0
B1 - 0
B2 - 1
The calculation converges smoothly to
Total Energy : -75.56349710 Eh -2056.18729 eV
With the final orbitals being:
SPIN UP ORBITALS
NO OCC E(Eh) E(eV) Irrep
0 1.0000 -21.127827 -574.9174 1-A1
1 1.0000 -1.867576 -50.8193 2-A1
2 1.0000 -1.192139 -32.4397 1-B2
3 1.0000 -1.124657 -30.6035 1-B1
4 1.0000 -1.085062 -29.5260 3-A1
5 0.0000 -0.153303 -4.1716 4-A1
6 0.0000 -0.071324 -1.9408 2-B2
...
SPIN DOWN ORBITALS
0 1.0000 -21.081198 -573.6486 1-A1
1 1.0000 -1.710193 -46.5367 2-A1
2 1.0000 -1.152855 -31.3708 1-B2
3 1.0000 -1.032556 -28.0973 1-B1
4 0.0000 -0.306683 -8.3453 3-A1
5 0.0000 -0.139418 -3.7937 4-A1
6 0.0000 -0.062261 -1.6942 2-B2
7 0.0000 0.374727 10.1968 3-B2
...
Suppose now that we want to converge on an excited state formed by flipping the spin-beta HOMO and
LUMO that have different symmetries.
! SVP UseSym
! moread
%moinp "Test-SYM-H2O+.gbw"
%scf rotate {3,4,90,1,1}
end
end
* xyz 1 2
O 0.000000 0.000000 0.068897
H 0.000000 0.788011 -0.546765
H 0.000000 -0.788011 -0.546765
*
The program now finds:

A1 - 3
A2 - 0
B1 - 1
B2 - 1
A1 - 2
A2 - 0
B1 - 1
B2 - 1
And converges smoothly to
Which is obviously an excited state of the H2 O+ molecule. In this situation (and in many others) it is an
advantage to have symmetry adapted orbitals.
8.1.2 MP2
8.1.2.1 MP2 and RI-MP2 Energies
You can do conventional or integral direct MP2 calculations for RHF, UHF or high-spin ROHF reference
wavefunctions. MP3 functionality is not implemented as part of the MP2 module, but can be accessed
through the MDCI module. Analytic gradients and Hessians are available for RHF and UHF. The frozen
core approximation is used by default. An extensive coverage of MP2 exists in the literature. [5972]
! MP2 RHF TZVP TightSCF

%mp2 MaxCore 100
end
%paras rCO = 1.20
ACOH = 120
rCH = 1.08
end
* int 0 1
C 0 0 0 0.00 0.0 0.00
O 1 0 0 {rCO} 0.0 0.00
H 1 2 0 {rCH} {ACOH} 0.00
H 1 2 3 {rCH} {ACOH} 180.00
*
NOTE:
There are two algorithms for MP2 calculations without the RI approximation. The first one uses
main memory as much as possible. The second one uses more disk space and is usually faster (in
particular, if you run the calculations in single precision using ! FLOAT, UCFLOAT or CFLOAT). The
memory algorithm is used using Q1Opt >0 and the disk based algorithm uses Q1Opt = -1. Gradients
are presently only available for the memory based algorithm.
The RI approximation to MP2 [6972] is fairly easy to use, too. It results in a tremendous speedup of the
calculation, while errors in energy differences are very small. For example, consider the same calculation as
before:
# only the auxiliary basis set TZV/C is added to

# the keyword line
#
! RI-MP2 RHF TZVP TZV/C TightSCF
%mp2 MaxCore 100
end
%paras rCO = 1.20
ACOH = 120
rCH = 1.08
end
* int 0 1
C 0 0 0 0.00 0.0 0.00
O 1 0 0 {rCO} 0.0 0.00
H 1 2 0 {rCH} {ACOH} 0.00
H 1 2 3 {rCH} {ACOH} 180.00
*
Generally, the RI approximation can be switched on by setting RI true in the %MP2 block. Specification of
an appropriate auxiliary basis set (/C) for correlated calculations is required. Note that if the RIJCOSX
method (section 8.1.4.3) or the RI-JK method (section 8.1.4.4) is used to accelerate the SCF calculation, then
two basis sets should be specified: firstly the appropriate Coulomb (/J) or exchange fitting set (/JK),
and secondly the correlation fitting set (/C), as shown in the example below.
# Simple input line for RIJCOSX:

! RHF RI-MP2 RIJCOSX def2-TZVP def2/J def2-TZVP/C TightSCF
# Simple input line for RI-KJ:

! RHF RI-MP2 RI-JK def2-TZVP def2/JK def2-TZVP/C TightSCF
The MP2 module can also do Grimmes spin-component scaled MP2 [73]. It is a semi-empirical modification
of MP2 which applies different scaling factors to same-spin and opposite-spin components of the MP2 energy.
Typically it gives a fair bit better results than MP2 itself.
#
# Spin-component scaled MP2 example
#
! SCS-MP2 RHF TZV(2d,2p) TightSCF
%paras rCO = 1.20
ACOH = 120
rCH = 1.08
end
* int 0 1
C 0 0 0 0.00 0.0 0.00
O 1 0 0 {rCO} 0.0 0.00
H 1 2 0 {rCH} {ACOH} 0.00
H 1 2 3 {rCH} {ACOH} 180.00
*
Energy differences with SCS-MP2 appear to be much better than from MP2 itself according to Grimmes
detailed evaluation study. For the sake of efficiency, it is beneficial to make use of the RI approximation using
the RI-SCS-MP2 keyword. The opposite-spin and same-spin scaling factors can be modified using PS and PT
in the MP2-block, respectively. By default, PS = 6/5 and PT = 1/3.
NOTE
In very large RI-MP2 runs you can cut down the amount of main memory used by a factor of two
if you use the keyword ! FLOAT. This is more important in gradient runs than in single point runs.
Deviations from double precision values for energies and gradients should be in the Eh and sub-Eh
range. However, we have met cases where this option introduced a large and unacceptable error, in
particular in transition metal calculations. You are therefore adviced to be careful and check things
out beforehand.
8.1.2.2 Frozen Core Options
In MP2 energy and gradient runs the Frozen Core (FC) approximation is applied by default. This implies that
the core electrons are not included in the perturbation treatment, since the inclusion of dynamic correlation
in the core electrons usually effects relative energies or geometry parameters insignificantly.
The frozen core option can be switched on or off with FrozenCore or NoFrozenCore in the simple input line.
Furthermore, frozen orbitals can be selected by means of an energy window:
%method FrozenCore FC_EWIN end

%mp2 ewin -1.5, 1.0e3 end
More information and the different options can be found in section 9.8
8.1.2.3 Orbital Optimized MP2 Methods
By making the Hylleraas functional stationary with respect to the orbital rotations one obtains the orbital-
optimized MP2 method that is implemented in ORCA in combination with the RI approximation (OO-
RI-MP2). One obtains from these calculations orbitals that are adjusted to the dynamic correlation field
at the level of second order many-body perturbation theory. Also, the total energy of the OO-RI-MP2
method is lower than that of the RI-MP2 method itself. One might think of this method as a special form of
multiconfigurational SCF theory except for the fact that the Hamiltonian is divided into a 0th order term
and a perturbation.
The main benefit of the OO-RI-MP2 method is that it repairs the poor HartreeFock orbitals to some
extent which should be particularly beneficial for systems which suffer from the inbalance in the Hartree-Fock
treatment of the Coulomb and the Exchange hole. Based on the experience gained so far, the OO-RI-MP2
method is no better than RI-MP2 itself for the thermochemistry of organic molecules. However, for reactions
barriers and radicals the benefits of OO-MP2 over MP2 are substantial. This is particularly true with respect
to the spin-component scaled variant of OO-RI-MP2 that is OO-RI-SCS-MP2. Furthermore, the OO-RI-MP2
method substantially reduces the spin contamination in UHF calculations on radicals.
Since every iteration of the OO-MP2 method is as expensive as a RI-MP2 relaxed density calculation, the
computational cost is much higher than for RI-MP2 itself. One should estimate about a factor of 10 increase
in computation time with respect to the RI-MP2 time of a normal calculation. This may still be feasible
for calculations in the range of 10002000 basis functions (the upper limit, however, implies very significant
computational costs). A full assessment of the orbital optimized MP2 method has been published. [74]
OO-RI-MP2 is triggered with ! OO-RI-MP2 or ! OO-RI-SCS-MP2 (with spin component scaling). The
method comes with new variables:
%mp2 OrbOpt true # turns on the orbital optimization

CalcS2 false # calculate the S**2 expectation value
# in spin-unrestricted calculations
MaxOrbIter 64 # Max. number of iterations
MP2Shift 0.1 # Level shift for the procedure

end
The solver is a simple DIIS type scheme with additional level shifting. We have found that it is not really
beneficial to first converge the Hartree-Fock equations. Thus it is sensible to additionally use the keyword !
noiter in order to turn off the standard Hartree-Fock SCF process before entering the orbital optimizations.
The OO-RI-MP2 method is implemented for RHF and UHF reference wavefunctions. Analytic gradients are
available.
The density does not need to be requested separately in OO-RI-MP2 calculations because it is automatically
calculated. Also, there is no distinction between relaxed and unrelaxed densities because the OO-RI-MP2
energy is fully stationary with respect to all wavefunction parameters and hence the unrelaxed and relaxed
densities coincide.
8.1.2.4 MP2 and RI-MP2 Gradients and Hessians
Geometry optimization with MP2, RI-MP2, SCS-MP2 and RI-SCS-MP2 proceeds just as with any SCF
method. Frequencies can be calculated analytically in all-electron calculations. With frozen core orbitals,
second derivatives of any kind are currently only available numerically. The RIJCOSX approximation (section
8.1.4.3) is supported in RI-MP2 and hence also in double-hybrid DFT gradient runs. This leads to large
speedups in larger calculations, particularly if the basis sets are accurate.
#
# MP2 optimization example
#
! RHF SCS-MP2 TZVP TZV/C TightSCF Opt NoFrozenCore
* int 0 1
C 0 0 0 0.00 0.0 0.00
O 1 0 0 1.20 0.0 0.00
H 1 2 0 1.09 120.0 0.00
H 1 2 3 1.09 120.0 180.00
*
This job results in:
------------------------------------------------------------------------
Redundant Internal Coordinates
--- Optimized Parameters ---

(Angstroem and degrees)
Definition OldVal dE/dq Step FinalVal

-------------------------------------------------------------------------
1. B(O 1,C 0) 1.2115 0.000515 -0.0003 1.2112
2. B(H 2,C 0) 1.1004 -0.000033 0.0001 1.1005

3. B(H 3,C 0) 1.1004 -0.000033 0.0001 1.1005
4. A(H 2,C 0,O 1) 121.93 0.000028 -0.00 121.93
5. A(H 3,C 0,O 1) 121.93 0.000028 -0.00 121.93
6. A(H 3,C 0,H 2) 116.14 -0.000056 0.01 116.14
7. I(O 1,H 3,H 2,C 0) -0.00 -0.000000 0.00 -0.00
-------------------------------------------------------------------------
Just to demonstrate the accuracy of RI-MP2, here is the result with RI-SCS-MP2 instead of SCS-MP2:
------------------------------------------------------------------------


-------------------------------------------------------------------------
1. B(O 1,C 0) 1.2115 0.000515 -0.0003 1.2112
2. B(H 2,C 0) 1.1004 -0.000033 0.0001 1.1005
3. B(H 3,C 0) 1.1004 -0.000033 0.0001 1.1005
4. A(H 2,C 0,O 1) 121.93 0.000028 -0.00 121.93
5. A(H 3,C 0,O 1) 121.93 0.000028 -0.00 121.93
6. A(H 3,C 0,H 2) 116.14 -0.000056 0.01 116.14
7. I(O 1,H 3,H 2,C 0) -0.00 -0.000000 0.00 0.00
-------------------------------------------------------------------------
You see that nothing is lost in the optimized geometry through the RI approximation thanks to the efficient
and accurate RI-auxiliary basis sets of the Karlsruhe group (in general the deviations in the geometries
between standard MP2 and RI-MP2 are very small). Thus, RI-MP2 really is a substantial improvement in
efficiency over standard MP2.
Geometric gradients and Hessians can be calculated with RI-MP2 in conjunction with the RIJCOSX method.
They are called the same way as with a conventional SCF wave function, for example to perform a geometry
optimization with tight convergence parameters:
! RI-MP2 def2-TZVPP def2/J def2-TZVPP/C TightSCF RIJCOSX

! TightOpt
...
8.1.2.5 MP2 Properties, Densities and Natural Orbitals
The MP2 method can be used to calculate electric and magnetic properties such as dipole moments,
polarizabilities, hyperfine couplings or g-tensors. For this purpose, the appropriate MP2 density needs to be
requested - otherwise the properties are calculated using the SCF density!
Two types of densities can be constructed - an unrelaxed density (which basically corresponds to the MP2
expectation value density) and a relaxed density which incorporates orbital relaxation. For both sets of
densities a population analysis is printed if the SCF calculation also requested this population analysis. These
two densities are stored as JobName.pmp2ur.tmp and JobName.pmp2re.tmp, respectively. The corresponding
spin densities are also constructed.
In addition to the density options, the user has the ability to construct MP2 natural orbitals. If relaxed
densities are available, the program uses the relaxed densities and otherwise the unrelaxed ones. The natural
orbitals are stored as JobName.mp2nat which is a GBW type file that can be read as input for other jobs (for
example, it is sensible to start CASSCF calculations from MP2 natural orbitals). The density construction
can be controlled separately in the input file (even without running a gradient or optimization) by:
#
# MP2 densities and natural orbitals
#
%mp2 Density none # no density
unrelaxed # unrelaxed density
relaxed # relaxed density
NatOrbs true # Natural orbital construction on or off
end
Below is a calculation of the dipole and quadrupole moments of a water molecule:
! RHF RI-MP2 def2-SVP def2-SVP/C

%mp2 density relaxed end
%elprop dipole true
quadrupole true
end
* int 0 1
O 0 0 0 0 0 0
H 1 0 0 0.9584 0 0
H 1 2 0 0.9584 104.45 0
*
Another example is a simple g-tensor calculation with MP2:
! UHF RI-MP2 def2-SVP def2-SVP/C TightSCF SOMF(1X) NoFrozenCore

%eprnmr gtensor 1
ori -3
end
%mp2 density relaxed end
* int 1 2
O 0 0 0 0 0 0
H 1 0 0 1.1056 0 0
H 1 2 0 1.1056 109.62 0
*
8.1.2.6 Explicitly correlated MP2 calculations
ORCA features an efficient explicit correlation module that is available for MP2 and coupled-cluster calculations
(section 8.1.3.6). It is described below in the context of coupled-cluster calculations.
8.1.2.7 Local MP2 calculations
Purely domain-based local MP2 methodology dates back to Pulay and has been developed further by Werner,
utz and co-workers. ORCA features a local MP2 method (DLPNO-MP2) that combines the ideas of
Sch
domains and local pair natural orbitals, so that RI-MP2 energies are reproduced efficiently to within chemical
accuracy. Due to the intricate connections with other DLPNO methods, reading of the sections 8.1.3.8 and
and 9.10.4 is recommended. A full description of the method for RHF reference wave functions has been
published. [75] Explicit correlation is supported through the DLPNO-MP2-F12 methodology. [76]
Since DLPNO-MP2 employs an auxiliary basis set to evaluate integrals, its energies converge systematically
to RI-MP2 as thresholds are tightened. The computational effort of DLPNO-MP2 with default settings is
usually comparable with or less than that of a Hartree-Fock calculation. However, for small and medium-sized
molecules, RI-MP2 is even faster than DLPNO-MP2.
Calculations on open-shell systems are supported through a UHF treatment. While most approximations
are consistent between the RHF and UHF versions, this is not true for the PNO spaces. DLPNO-MP2
gives different energies for closed-shell molecules in the RHF and UHF formalisms. Therefore,
numbers obtained through the UHF formalism must only be compared to other UHF numbers
(and not to RHF numbers), even for closed-shell species. As for canonical MP2, ROHF reference
wave functions are subject to an ROMP2 treatment through the UHF machinery. It is not consistent with
the RHF version of DLPNO-MP2, unlike in the case of RHF-/ROHF-DLPNO-CCSD.
Input for DLPNO-MP2 requires little specification from the user:
# DLPNO-MP2 calculation with standard settings

# sufficient for most purposes
! def2-TZVP def2-TZVP/C DLPNO-MP2 TightSCF
# OR: DLPNO-MP2 with tighter thresholds

# May be interesting for weak interactions, calculations with diffuse basis sets etc.
! def2-TZVP def2-TZVP/C DLPNO-MP2 TightPNO TightSCF
%maxcore 2000
*xyz 0 1
... (coordinates)
*
Noteworthy aspects of the DLPNO-MP2 method:

Both DLPNO-CCSD(T) and DLPNO-MP2 are linear-scaling methods (albeit the former has a larger
prefactor). This means that if a DLPNO-MP2 calculation can be performed, DLPNO-CCSD(T) is
often going to be within reach, too. However, CCSD(T) is generally much more accurate than MP2
and thus should be given preference.
A correlation fitting set must be provided, as the method makes use of the RI approximation.
Canonical RI-MP2 energy differences are typically reproduced to within a fraction of 1 kcal/mol. The
default thresholds have been chosen so as to reproduce about 99.9 % of the total RI-MP2 correlation
energy.
The preferred way to control the accuracy of the method is by means of specifying LoosePNO,
NormalPNO and TightPNO keywords. NormalPNO corresponds to default settings and does not
need to be given explicitly. More details and an exhaustive list of input parameters are provided in
section 9.9.7. Note that the thresholds differ from DLPNO coupled cluster.
Results obtained from RI-MP2 and DLPNO-MP2, or from DLPNO-MP2 with different accuracy
settings, must never be mixed, such as when computing energy differences. In calculations involving
open-shell species, even the closed-shell molecules need to be subject to a UHF treatment.
Spin-component scaled DLPNO-MP2 calculations are invoked by using the ! DLPNO-SCS-MP2 keyword
instead of ! DLPNO-MP2 in the simple input line. Weights for same-spin and opposite-spin contributions
can be adjusted as described for the canonical SCS-MP2 method.
The frozen core approximation is used by default. If core orbitals are involved in the calculation, they
are subject to the treatment described in section 9.9.7.
Calculations can be performed in parallel.
No analytic derivatives are implemented.
It may be beneficial to accelerate the Hartree-Fock calculation by means of the RIJCOSX method
(requiring specification of a second auxiliary set).
Explicit correlation has been implemented in the DLPNO-MP2-F12 methodology for RHF reference wave
functions. The available approaches are C (keyword ! DLPNO-MP2-F12) and the somewhat more approximate
D (keyword ! DLPNO-MP2-F12/D). Approach D is generally recommended as it results in a significant speedup
while leading only to small errors relative to approach C. In addition to the MO and correlation fitting sets,
a CABS basis set is also required for both F12 approaches as shown below.
# DLPNO-MP2-F12 calculation using approach C

! cc-pVDZ-F12 aug-cc-pVDZ/C cc-pVDZ-F12-CABS DLPNO-MP2-F12 TightSCF
# OR: DLPNO-MP2-F12 calculation using approach D (recommended)

! cc-pVDZ-F12 aug-cc-pVDZ/C cc-pVDZ-F12-CABS DLPNO-MP2-F12/D TightSCF
8.1.3 Coupled-Cluster and Coupled-Pair Methods
8.1.3.1 Basics
The coupled-cluster method is presently available for RHF and UHF references. The implementation is fairly
efficient and suitable for large-scale calculations. The most elementary use of this module is fairly simple.
! METHOD
# where METHOD is:
# CCSD CCSD(T) QCISD QCISD(T) CPF/n NCPF/n CEPA/n NCEPA/n
# (n=1,2,3 for all variants) ACPF NACPF AQCC CISD
! AOX-METHOD
# computes contributions from integrals with 3- and 4-external
# labels directly from AO integrals that are pre-stored in a
# packed format suitable for efficient processing
! AO-METHOD
# computes contributions from integrals with 3- and 4-external
# labels directly from AO integrals. Can be done for integral
# direct and conventional runs. In particular, the conventional
# calculations can be very efficient
! MO-METHOD (this is the default)

# performs a full four index integral transformation. This is
# also often a good choice
! RI-METHOD
# selects the RI approximation for all integrals. Rarely advisable
! RI34-METHOD
# selects the RI approximation for the integrals with 3- and 4-
# external labels
#
# The module has many additional options that are documented
# later in the manual.
! RCSinglesFock
! RIJKSinglesFock
! NoRCSinglesFock
! NoRIJKSinglesFock
# Keywords to select the way the so-called singles Fock calculation
# is evaluated. The first two keywords turn on, the second two turn off
# RIJCOSX or RIJK, respectively.
NOTE
The same FrozenCore options as for MP2 are applied in the MDCI module.
The computational effort for these methods is high O(N6 ) for all methods and O(N7 ) if the triples
correction is to be computed (calculations based on an unrestricted determinant are roughly 3 times more
expensive than closed-shell calculations and approximately six times more expensive if triple excitations
are to be calculated). This restricts the calculations somewhat: on presently available PCs 300400 basis
functions are feasible and if you are patient and stretch it to the limit it may be possible to go up to 500600;
if not too many electrons are correlated may be even up to 800900 basis functions (then using AO-direct
methods).
TIP
For calculations on small molecules and large basis sets the MO-METHOD option is usually the most
efficient; say perhaps up to about 300 basis functions. For integral conventional runs the AO-METHOD
may even more efficient.
For large calculations (>300 basis functions) the AO-METHOD option is a good choice. If, however,
you use very deeply contracted basis sets such as ANOs these calculations should be run in the integral
conventional mode.
AOX-METHOD is usually slightly less efficient than MO-METHOD or AO-METHOD.
RI-METHOD is seldom the most efficient choice. If the integral transformation time is an issue than
you can select %mdci trafotype trafo ri or choose RI-METHOD and then %mdci kcopt kc ao.
Regarding the singles Fock keywords (RCSinglesFock, etc.), the program usually decides which method
to use to evaluate the singles Fock term. For more details on the nature of this term, and options
related to its evaluation, see 9.10.6.
To put this into perspective, consider a calculation on serine with the cc-pVDZ basis set a basis on the
lower end of what it suitable for a highly correlated calculation. The time required to solve the equations is
listed in Table 8.1. We can draw the following conclusions:
As long as one can store the integrals and the I/O system of the computer is not the bottleneck, the
most efficient way to do coupled-cluster type calculations is usually to go via the full transformation
(it scales as O(N5 ) whereas the later steps scale as O(N6 ) and O(N7 ) respectively).
AO-based coupled-cluster calculations are not much inferior. For larger basis sets (i.e. when the ratio
of virtual to occupied orbitals is larger), the computation times will be even more favorable for the
AO based implementation. The AO direct method uses much less disk space. However, when you use
a very expensive basis set the overhead will be larger than what is observed in this example. Hence,
conventionally stored integrals if affordable are a good choice.
AOX based calculations runs at essentially the same speed as AO based calculations. Since AOX
based calculations take four times as much disk space they are pretty much outdated and the AOX
implementation is only kept for historical reasons.
RI based coupled-cluster methods are significantly slower. There are some disk space savings but the
computationally dominant steps are executed less efficiently.
CCSD is at most 10% more expensive than QCISD. With the latest AO implementation the awkward
coupled-cluster terms are handled efficiently.
CEPA is not much more than 20% faster than CCSD. In many cases CEPA results wil be better than
CCSD and then it is a real saving compared to CCSD(T), which is the most rigorous.
If triples are included practically the same comments apply for MO versus AO based implementations
as in the case of CCSD.
ORCA is quite efficient in this type of calculation but it is also clear that the range of application of these
rigorous methods is limited as long as one uses canonical MOs. ORCA implements novel variants of the
so-called local Coupled-Cluster method which can calculate large, real-life molecules in a linear scaling time.
This will be addressed in Sec. 8.1.3.8.
Table 8.1: Computer times (minutes) for solving the coupled-cluster/coupled-pair equations for Serine (cc-
pVDZ basis set).
Method SCFMode Time (min)
MO-CCSD Conv 38.2
AO-CCSD Conv 47.5
AO-CCSD Direct 50.8
AOX-CCSD Conv 48.7
RI-CCSD Conv 64.3
AO-QCISD Conv 44.8
AO-CEPA/1 Conv 40.5
MO-CCSD(T) Conv 147.0
AO-CCSD(T) Conv 156.7
All of these methods are designed to cover dynamic correlation in systems where the Hartree-Fock determinant
dominates the wavefunctions. The least attractive of these methods is CISD which is not size-consistent and
therefore practically useless. The most rigorous are CCSD(T) and QCISD(T). The former is perhaps to
be preferred since it is more stable in difficult situations.1 One can get highly accurate results from such
calculations. However, one only gets this accuracy in conjunction with large basis sets. It is perhaps not
very meaningful to perform a CCSD(T) calculation with a double-zeta basis set (see Table 8.2). The very
least basis set quality required for meaningful results would perhaps be something like def2-TZVP(-f) or
preferably def2-TZVPP (cc-pVTZ, ano-pVTZ). For accurate results quadruple-zeta and even larger basis sets
are required and at this stage the method is restricted to rather small systems.
Let us look at the case of the potential energy surface of the N2 molecule. We study it with three different
basis sets: TZVP, TZVPP and QZVP. The input is the following:
1
The exponential of the T1 operator serves to essentially fully relax the orbitals of the reference wavefunction. This
is not included in the QCISD model that only features at most a blinear T1T2 term in the singles residuum.
Hence, if the Hartree-Fock wavefunction is a poor starting point but static correlation is not the main problem,
CCSD is much preferred over QCISD. This is not uncommon in transition metal complexes.
! RHF TZVPP CCSD(T)

%paras R= 1.05,1.13,8
end
* xyz 0 1
N 0 0 0
N 0 0 {R}
*
For even higher accuracy we would need to introduce relativistic effects and - in particular - turn the core
correlation on. 2
Table 8.2: Computed spectroscopic constants of N2 with coupled-cluster methods.

Method Basis set Re (pm) e (cm1 ) e xe (cm1 )
CCSD(T) SVP 111.2 2397 14.4
TZVP 110.5 2354 14.9
TZVPP 110.2 2349 14.1
QZVP 110.0 2357 14.3
ano-pVDZ 111.3 2320 14.9
ano-pVTZ 110.5 2337 14.4
ano-pVQZ 110.1 2351 14.5
CCSD QZVP 109.3 2437 13.5
Exp 109.7 2358.57 14.32
One can see from Table 8.2 that for high accuracy - in particular for the vibrational frequency - one needs
both - the connected triple-excitations and large basis sets (the TZVP result is fortuitously good). While this
is an isolated example, the conclusion holds more generally. If one pushes it, CCSD(T) has an accuracy (for
reasonably well-behaved systems) of approximately 0.2 pm in distances, <10 cm1 for harmonic frequencies
and a few kcal/mol for atomization energies.3 It is also astonishing how well the Ahlrichs basis sets do in
these calculations even slightly better than the much more elaborate ANO bases.
NOTE:
The quality of a given calculation is not always high because it carries the label Coupled-Cluster.
Accurate results are only obtained in conjunction with large basis sets and for systems where the HF
approximation is a good 0th order starting point.
2
Note that core correlation is not simply introduced by including the core orbitals in the correlation problem. In
addition, special correlation core-polarization functions are needed. They have been standardized for a few elements
in the cc-pCVxZ (X=D,T,Q,5,6) basis sets.
3
However, in recent years it became more evident that even CCSD(T) achieves its high apparent accuracy through
error cancellations. The full CCSDT method (triples fully included) usually performs worse than CCSD(T). The
reason is that the (T) correction undershoots the effects of the triples to some extent and thereby compensates for
the neglect of connected quadruple excitations. For very high accuracy quantum chemistry, even these must be
considered. The prospects for treating chemically more relevant molecules with such methods is not particularly
bright for the foreseeable future. . .
8.1.3.2 Coupled-Cluster Densities
If one is mainly accustomed to Hartree-Fock or DFT calculations, the calculation of the density matrix
is more or less a triviality and is automatically done together with the solution of the self-consistent field
equations. Unfortunately, this is not the case in coupled-cluster theory (and also not in MP2 theory). The

underlying reason is that in coupled-cluster theory, the expansion of the exponential eT in the expectation
value

h|Epq |i heT 0 |Epq |eT 0 i
Dpq = =
h|i heT 0 |eT 0 i
only terminates if all possible excitation levels are exhausted, i.e., if all electrons in the reference determinant
0 (typically the HF determinant) are excited from the space of occupied to the space of virtual orbitals
(here Dpq denotes the first order density matrix, Epq are the spin traced second quantized orbital replacement
operators, and T is the cluster operator). Hence, the straightforward application of these equations is far
too expensive. It is, however, possible to expand the exponentials and only keep the linear term. This
then defines a linearized density which coincides with the density that one would calculate from linearized
coupled-cluster theory (CEPA/0). The difference to the CEPA/0 density is that converged coupled-cluster
amplitudes are used for its evaluation. This density is straightforward to compute and the computational
effort for the evaluation is very low. Hence, this is a density that can be easily produced in a coupled-cluster
run. It is not, however, what coupled-cluster aficionados would accept as a density.
The subject of a density in coupled-cluster theory is approached from the viewpoint of response theory.
Imagine one adds a perturbation of the form
X
H () = hpq Epq
pq
to the Hamiltonian. Then it is always possible to cast the first derivative of the total energy in the form:
dE X
(response)
= Dpq hpq
d pq
(response)
This is a nice result. The quantity Dpq is the so-called response density. In the case of CC theory
where the energy is not obtained by variational optimization of an energy functional, the energy has to be
replaced by a Lagrangian reading as follows:
X 1X X
T |0 i
LCC = EHF + zai Fai + hij||abitij
ab + z h|He
ai
4
ijab
Here denotes any excited determinant (singly, doubly, triply, ....). There are two sets of Lagrange multipliers:
the quantities zai that guarantee that the perturbed wavefunction fulfills the Hartree-Fock conditions by
making the off-diagonal Fock matrix blocks zero and the quantities z that guarantee that the coupled-cluster
projection equations for the amplitudes are fulfilled. If both sets of conditions are fulfilled then the coupled-
cluster Lagrangian simply evaluates to the coupled-cluster energy. The coupled-cluster Lagrangian can be
made stationary with respect to the Lagrangian multipliers zai and z . The response density is then defined
through:
dLCC X
(response)
= Dpq hpq
d pq
The density Dpq appearing in this equation does not have the same properties as the density that would arise
from an expectation value. For example, the response density can have eigenvalues lower than 0 or larger
than 2. In practice, the response density is, however, the best density there is for coupled-cluster theory.
Unfortunately, the calculation of the coupled-cluster response density is quite involved because additional sets
of equations need to be solved in order to determine the zai and z . If only the equations for z are solved one
speaks of an unrelaxed coupled-cluster density. If both sets of equations are solved, one speaks of a relaxed
coupled-cluster density. For most intents and purposes, the orbital relaxation effects incorporated into the
relaxed density are small for a coupled-cluster density. This is so, because the coupled-cluster equations

contain the exponential of the single excitation operator eT1 = exp( ai tia Eia ). This brings in most of the
P
effects of orbital relaxation. In fact, replacing the T1 operator by the operator = ai ia (Eia Eai ) would
P
provide all of the orbital relaxation thus leading to orbital optimized coupled-cluster theory (OOCC).
Not surprisingly, the equations that determine the coefficients z (the Z vector equations) are as complicated
as the coupled-cluster amplitude equations themselves. Hence, the calculation of the unrelaxed coupled-cluster
density matrix is about twice as expensive as the calculation of the coupled-cluster energy (but not quite as
with proper program organization terms can be reused and the Z vector equations are linear equations that
converge somewhat better than the non-linear amplitude equations).
ORCA features the calculation of the unrelaxed coupled-cluster density on the basis of the Z vector equations
for closed- and open-shell systems. If a fully relaxed coupled-cluster density is desired then ORCA still
features the orbital-optimized coupled-cluster doubles method (OOCCD). This is not exactly equivalent to
the fully relaxed CCSD density matrix because of the operator instead of T1 . However, results are very
close and orbital optimized coupled-cluster doubles is the method of choice if orbital relaxation effects are
presumed to be large.
In terms of ORCA keywords, the coupled-cluster density is obtained through the following keywords:
#
# coupled-cluster density
#
%mdci density none
linearized
unrelaxed
orbopt
end
which will work together with CCSD or QCISD (QCISD and CCSD are identical in the case of OOCCD
because of the absence of single excitations). Note, that an unrelaxed density for CCSD(T) is NOT available.
Instead of using the density option orbopt in the mdci-block, OOCCD can also be invoked by using the
keyword:
! OOCD
8.1.3.3 Static versus Dynamic Correlation
Having said that, let us look at an abuse of the single reference correlation methods by studying (very
superficially) a system which is not well described by a single HF determinant. This already occurs for the
twisting of the double bond of C2 H4 . At a 90 twist angle the system behaves like a diradical and should be
described by a multireference method (see section 8.1.7)
Figure 8.1: A rigid scan along the twisting coordinate of C2 H4 . The inset shows the T1 diagnostic
for the CCSD calculation.
As can be seen in Figure 8.1, there is a steep rise in energy as one approaches a 90 twist angle. The HF
curve is actually discontinuous and has a cusp at 90 . This is immediately fixed by a simple CASSCF(2,2)
calculation which gives a smooth potential energy surface. Dynamic correlation is treated on top of the
CASSCF(2,2) method with the MRACPF approach as follows:
#
# twisting the double bond of C2H4
#
! RHF SV(P) SV/C SmallPrint NoPop MRACPF
%casscf nel 2
norb 2
mult 1
nroots 1
TrafoStep RI
end
%mrci tsel 1e-10
tpre 1e-10
end
%method scanguess pmodel
end
%paras R= 1.3385
Alpha=0,180,18
end
* int 0 1
C 0 0 0 0 0 0
C 1 0 0 {R} 0 0
H 1 2 0 1.07 120 0
H 1 2 3 1.07 120 180
H 2 1 3 1.07 120 {Alpha}
H 2 1 3 1.07 120 {Alpha+180}
*
This is the reference calculation for this problem. One can see that the RHF curve is far from the MRACPF
reference but the CASSCF calculation is very close. Thus, dynamic correlation is not important for this
problem! It only appears to be important since the RHF determinant is such a poor choice. The MP2
correlation energy is insufficient in order to repair the RHF result. The CCSD method is better but still falls
short of quantitative accuracy. Finally, the CCSD(T) curve is very close the MRACPF. This even holds for
the total energy (inset of Figure 8.2) which does not deviate by more than 23 mEh from each other. Thus,
in this case one uses the powerful CCSD(T) method in an inappropriate way in order to describe a system
that has multireference character. Nevertheless, the success of CCSD(T) shows how stable this method is
even in tricky situations. The alarm bell for CCSD and CCSD(T) is the so-called T1 -diagnostic4 that is
also shown in Figure 8.2. A rule of thumb says, that for a value of the diagnostic of larger than 0.02 the
results are not to be trusted. In this calculation we have not quite reached this critical point although the T1
diagnostic blows up around the 90 twist.
Figure 8.2: Comparison of the CCSD(T) and MRACPF total energies of the C2 H4 along the twist-
ing coordinate. The inset shows the difference E(MRACPF)-E(CCSD(T)).
The computational cost (disregarding the triples) is such that the CCSD method is the most expensive
followed by QCISD (10% cheaper) and all other methods (about 50% to a factor of two cheaper than
CCSD). The most accurate method is generally CCSD(T). However, this is not so clear if the triples are
4
It is defined as kT1 k /N 1/2 where T1 are the singles amplitudes and N the number of correlated electrons. The
original reference is [77]
omitted and in this regime the coupled pair methods (in particular CPF/1 and NCPF/15 ) can compete with
CCSD.
Let us look at the same type of situation from a slightly different perspective and dissociate the single bond
of F2 . As is well known, the RHF approximation fails completely for this molecule and predicts it to be
unbound. Again we use a much too small basis set for quantitative results but it is enough to illustrate the
principle.
We first generate a reference PES with the MRACPF method:
! RHF SVP SV/C MRACPF

%casscf nel 2
norb 2
nroots 1
mult 1
end
%mrci tsel 1e-10
tpre 1e-10
end
%paras R= 3.0,1.3,35
end
* xyz 0 1
F 0 0 0
F 0 0 {R}
*
Note that we scan from outward to inward. This helps the program to find the correct potential energy
surface since at large distances the and orbitals are close in energy and fall within the desired 2 2
window for the CASSCF calculation (see section 8.1.7). Comparing the MRACPF and CASSCF curves it
becomes evident that the dynamic correlation brought in by the MRACPF procedure is very important and
changes the asymptote (loosely speaking the binding energy) by almost a factor of two. Around the minimum
(roughly up to 2.0 A) the CCSD(T) and MRACPF curves agree beautifully and are almost indistinguishable.
Beyond this distance the CCSD(T) calculation begins to diverge and shows an unphysical behavior while
the multireference method is able to describe the entire PES up to the dissociation limit. The CCSD curve
is qualitatively ok but has pronounced quantitative shortcomings: it predicts a minimum that is much too
short and a dissociation energy that is much too high. Thus, already for this rather simple molecule, the
effect of the connected triple excitations is very important. Given this (rather unpleasant) situation, the
behavior of the much simpler CEPA method is rather satisfying since it predicts a minimum and dissociation
energy that is much closer to the reference MRACPF result than CCSD or CASSCF. It appears that in this
particular case CEPA/1 and CEPA/2 bracket the correct result.
As for MP2 calculations, the RI approximation can be introduced. It does not lead to spectacular performance
gains but easens the burden for the integral transformation in larger calculations. The error introduced by
the RI-approximation is usually negligible. For larger systems the AO or AOX methods are usually used.
5
The N methods have been suggested by [78] and are exclusive to ORCA. Please note that our NCPF/1 is different
from the MCPF method in the literature [79]. The original CPF method which we prefer is from [80]; see
also [81] for a nice review about the coupled pair approaches and the underlying philosophy.
Figure 8.3: Potential energy surface of the F2 molecule calculated with some single-reference meth-
ods and compared to the MRACPF reference.
8.1.3.4 Basis Sets for Correlated Calculations. The case of ANOs.
In HF and DFT calculations the generation and digestion of the two-electron repulsion integrals is usually
the most expensive step of the entire calculation. Therefore, the most efficient approach is to use loosely
contracted basis sets with as few primitives as possible the Ahlrichs basis sets (SVP, TZVP, TZVPP, QZVP,
def2-TZVPP, def2-QZVPP) are probably the best in this respect. Alternatively, the polarization-consistent
basis sets pc-1 through pc-4 could be used, but they are only available for H-Ar. For large molecules such
basis sets also lead to efficient prescreening and consequently efficient calculations.
This situation is different in highly correlated calculations such as CCSD and CCSD(T) where the effort
scales steeply with the number of basis functions. In addition, the calculations are usually only feasible for a
limited number of basis functions and are often run in the integral conventional mode since high angular
momentum basis functions are present and these are expensive to recomputed all the time. Hence, a different
strategy concerning the basis set design seems logical. It would be good to use as few basis functions as
possible but make them as accurate as possible. This is compatible with the philosophy of atomic natural
orbital (ANO) basis sets. Such basis sets are generated from correlated atomic calculations and replicate
the primitives of a given angular momentum for each basis function. Therefore, these basis sets are deeply
contracted and expensive but the natural atomic orbitals form a beautiful basis for molecular calculations.
In ORCA an accurate and systematic set of ANOs (ano-pVnZ, n = D, T, Q, 5 is incorporated). A related
strategy underlies the design of the correlation-consistent basis sets (cc-pVnZ, n = D, T, Q, 5, 6,. . . ) that are
also generally contracted except for the outermost primitives of the principal orbitals and the polarization
functions that are left uncontracted.
Let us study this subject in some detail using the H2 CO molecule at a standard geometry and compute the
SCF and correlation energies with various basis sets. In judging the results one should view the total energy
in conjunction with the number of basis functions and the total time elapsed. Looking at the data in the
Table below, it is obvious that the by far lowest SCF energies for a given cardinal number (2 for double-zeta, 3
for triple zeta and 4 for quadruple-zeta) are provided by the ANO basis sets. Using specially optimized ANO
integrals that is available since ORCA 2.7.0, the calculations are not even much more expensive than those
with standard basis sets. Obviously, the correlation energies delievered by the ANO bases are also the best of
all 12 basis sets tested. Hence, ANO basis sets are a very good choice for highly correlated calculations. The
advantages are particularly large for the early members (DZ/TZ).
Table 8.3: Comparison of various basis sets for highly correlated calculations
No. Basis
Basis set E(SCF) EC (CCSD(T)) Etot (CCSD(T)) Total Time
Fcns
cc-pVDZ 38 -113.876184 -0.34117952 -114.217364 2
cc-pVTZ 88 -113.911871 -0.42135475 -114.333226 40
cc-pVQZ 170 -113.920926 -0.44760332 -114.368529 695
def2-SVP 38 -113.778427 -0.34056109 -114.118988 2
def2- 90 -113.917271 -0.41990287 -114.337174 46
TZVPP
def2- 174 -113.922738 -0.44643753 -114.369175 730
QZVPP
pc-1 38 -113.840092 -0.33918253 -114.179274 2
pc-2 88 -113.914256 -0.41321906 -114.327475 43
pc-3 196 -113.922543 -0.44911659 -114.371660 1176
ano-pVDZ 38 -113.910571 -0.35822337 -114.268795 12
ano-pVTZ 88 -113.920389 -0.42772994 -114.348119 113
ano-pVQZ 170 -113.922788 -0.44995355 -114.372742 960
Figure 8.4: Error in mEh for various basis sets for highly correlated calculations relative to the
ano-pVQZ basis set.
Let us look at one more example in Table 8.4: the optimized structure of the N2 molecule as a function of
basis set using the MP2 method (these calculations are a bit older from the time when the ano-pVnZ basis
sets did not yet exist. Today, the ano-pVnZ would be prefered).
The highest quality basis set here is QZVP and it also gives the lowest total energy. However, this basis set
contains up to g-functions and is very expensive. The Bonn-ANO-TZ3P is of the same size as TZVPP and
gives the same result as QZVP for the geometry and an energy that is intermediate between TZVPP and
QZVP. To not use a set of f-functions has still a noticeable effect on the outcome of the calculations and leads
to an overestimation of the bond distance of 0.2 pm a small change but for benchmark calculations of this
kind still significant. Among these spd-only basis sets the Bonn-ANO-TZ2P basis set gives a better result
than TZV(2d,2p) and a lower energy as well. In fact, similarly as for the Bonn-ANO-TZ3P, the distances is as
good as that from TZVPP and the energy is intermediate between TZV(2d,2p) and TZVPP(TZV(2df,2pd)).
The error made by the TZVP basis set that lacks the second set of d-functions is surprisingly small even
though the deletion of the second d-set costs more than 20 mEh in the total energy.
A significant error on the order of 1 2 pm in the calculated distances is produced by smaller DZP type basis
sets which underlines once more that such basis sets are really too small for correlated molecular calculations
the ANO-DZP basis sets are too strongly biased towards the atom and the usual molecule targeted
DZP basis sets like SVP have the d-set designed to cover polarization but not correlation (the correlating
d-functions are steeper than the polarizing ones). Among the three tested basis sets the Bonn-ANO-DZP
fares best in this test and cc-pVDZ fares worst. The relatively good energy of the Bonn-ANO-DZP basis
certainly comes from the good description of the atoms. The performance of the very economical SVP basis
set should be considered as very good.
Essentially the same picture is obtained by looking at the (uncorrected for ZPE) binding energy calculated
at the MP2 level the largest basis set, QZVP gives the largest binding energy while the small basis set
underestimate it. The error of the DZP type basis sets is fairly large ( 2 eV) and therefore caution is
advisable when using such bases. In all cases it was found that the Bonn-ANO bases do slightly better than
the segmented contracted basis sets of the same size. This still holds for the calculated ionization potential of
the nitrogen atom. In principle, this is a worst case scenario for the ANO basis sets since they are supposedly
strongly biased towards the neutral atom. Yet, they fare no worse than the segmented contracted basis sets.
Here, an error cancellation is likely: The ANO bases recover more correlation (larger for the neutral) but are
biased towards the neutral (underestimating the SCF energy for the cation). This bias perhaps shows up
most clearly for the ANO-DZP basis set which gives a calculated IP that is accidentally almost right. The
largest ANO-TZ3P even slightly overshoots relative to the QZVP basis set, which is expected.
Table 8.4: Comparison of various basis sets for correlated calculations.

Basis set Req (pm) E(2N-N2 ) (eV) IP(N/N+ ) (eV) E(MP2) (Eh)
SVP 112.2 -9.67 14.45 -109.1677
cc-pVDZ 112.9 9.35 14.35 -109.2672
Bonn-ANO-DZP 112.1 9.45 14.58 -109.3098
TZVP 111.5 10.41 14.37 -109.3423
TZV(2d,2p) 111.4 10.61 14.49 -109.3683
Bonn-ANO-TZ2P 111.1 10.80 14.56 -109.3791
TZVPP 111.1 10.94 14.56 -109.3973
Bonn-ANO-TZ3P 110.9 11.18 14.65 -109.4108
QZVP 110.9 11.52 14.60 -109.4389
8.1.3.5 Automatic exptrapolation to the basis set limit
As eluded to in the previous section, one of the biggest problems with correlation calculations is the slow
convergence to the basis set limit. One possibility to overcome this problem is the use of explicitly correlated
methods. The other possibility is to use basis set extrapolation techniques. Since this involves some fairly
repetitive work, some procedures were hardwired into the ORCA program. So far, only energies are supported.
For extrapolation, a systematic series of basis sets is required. This is, for example, provided by the cc-pVnZ,
aug-cc-pVnZ or the corresponding ANO basis sets. Here n is the cardinal number that is 2 for the
double-zeta basis sets, 3 for triple-zeta, etc.
The convergence of the HF energy to the basis set limit is assumed to be given by:

(X) ()
ESCF = ESCF + A exp X (8.1)
(X) ()
Here, ESCF is the SCF energy calculated with the basis set with cardinal number X, ESCF is the basis set
limit SCF energy and A and are constants. The approach taken in ORCA is to do a two-point extrapolation.
This means that either A or have to be known. Here, we take A as to be determined and as a basis set
specific constant.
The correlation energy is supposed to converge as:
(X) (Y )
() X Ecorr Y Ecorr
Ecorr = (8.2)
X Y
The theoretical value for is 3.0. However, it was found by Truhlar and confirmed by us, that for 2/3
extrapolations = 2.4 performs considerably better.
For a number of basis sets, we have determined the optimum values for and [82]:
23 23 34 34
cc-pVnZ 4.42 2.46 5.46 3.05
pc-n 7.02 2.01 9.78 4.09
def2 10.39 2.40 7.88 2.97
ano-pVnZ 5.41 2.43 4.48 2.97
saug-ano-pVnZ 5.48 2.21 4.18 2.83
aug-ano-pVnZ 5.12 2.41
Since the values for 2/3 are close to 2.4, we always take this value. Likewise, all 3/4 and higher extrapolations
are done with = 3. However, the optimized values for are taken throughout.
Using the keyword ! Extrapolate(X/Y,basis), where X and Y are the corresponding successive cardinal
numbers and basis is the type of basis set requested (= cc, aug-cc, cc-core, ano, saug-ano, aug-ano,
def2) ORCA will calculate the SCF and optionally the MP2 or MDCI energies with two basis sets and
separately extrapolate.
The keyword works also in the following way: ! Extrapolate(n,basis) where n is the is the number of
energies to be used. In this way the program will start from a double-zeta basis and perform calculations
with n cardinal numbes and then extrapolates the different pairs of basis sets. Thus for example the keyword
! Extrapolate(3,CC) will perform calculations with cc-pVDZ, cc-pVTZ and cc-pVQZ basis sets and then
estimate the extrapolation results of both cc-pVDZ/cc-pVTZ and cc-pVTZ/cc-pVQZ combinations.
Let us take the example of the H2O molecule at the B3LYP/TZVP optimized geometry. The reference values
have been determined from a HF calculation with the decontracted aug-cc-pV6Z basis set and the correlation
energy was obtained from the cc-pV5Z/cc-pV6Z extrapolation. This gives:
E(SCF,CBS) = -76.066958 Eh
EC(CCSD(T),CBS) = -0.30866 Eh
Etot(CCSD(T),CBS) = -76.37561 Eh
Now we can see what extrapolation can bring in:
!CCSD(T) Extrapolate(2/3) TightSCF Conv Bohrs

* int 0 1
O 0 0 0 0 0 0
H 1 0 0 1.81975 0 0
H 1 2 0 1.81975 105.237 0
*
NOTE:
The RI-JK and RIJCOSX approximations work well together with this option and RI-MP2 is also
possible. Auxiliary basis sets are automatically chosen and can not be changed.
All other basis set choices, externally defined bases etc. will be ignored the automatic procedure
only works with the default basis sets!
The basis sets with the core postfix contain core correlation functions. By default it is assumed
that this means that the core electrons are also to be correlated and the frozen core approximation is
turned off. However, this can be overriden in the method block by choosing, e.g. %method frozencore
fc electrons end!
So far, the extrapolation is only implemented for single points and not for gradients. Hence, geometry
optimizations cannot be done in this way.
The extrapolation method should only be used with verytight SCF. For open shell methods, additional
caution is advised.
This gives:
Alpha(2/3) : 4.420 (SCF Extrapolation)

Beta(2/3) : 2.460 (correlation extrapolation)
SCF energy with basis cc-pVDZ: -76.026430944

SCF energy with basis cc-pVTZ: -76.056728252
Extrapolated CBS SCF energy (2/3) : -76.066581429 (-0.009853177)
MDCI energy with basis cc-pVDZ: -0.214591061

MDCI energy with basis cc-pVTZ: -0.275383015
Extrapolated CBS correlation energy (2/3) : -0.310905962 (-0.035522947)
Estimated CBS total energy (2/3) : -76.377487391
Thus, the error in the total energy is indeed strongly reduced. Let us look at the more rigorous 3/4
extrapolation:
Alpha(3/4) : 5.460 (SCF Extrapolation)

Beta(3/4) : 3.050 (correlation extrapolation)

SCF energy with basis cc-pVQZ: -76.064381269
Extrapolated CBS SCF energy (3/4) : -76.066687152 (-0.002305884)

MDCI energy with basis cc-pVQZ: -0.295324345
Extrapolated CBS correlation energy (3/4) : -0.309520369 (-0.014196024)
In our experience, the ANO basis sets extrapolate similarly to the cc-basis sets. Hence, repeating the entire
calculation with Extrapolate(3,ANO) gives:

Which is within 1 mEh of the estimated CCSD(T) basis set limit energy in the case of the 3/4 extrapolation
and within 2 mEh for the 2/3 extrapolation.
For larger molecules, the bottleneck of the calculation will be the CCSD(T) calculation with the larger basis
set. In order to avoid this expensive (or prohibitive) calculation, it is possible to estimate the CCSD(T)
energy at the basis set limit as:
(CCSD(T);Y ) (CCSD(T);X) (MP2;) (MP2;X)

Ecorr Ecorr + Ecorr Ecorr (8.3)
This assumes that the basis set dependence of MP2 and CCSD(T) is similar. One can then extrapolate as
before. Alternatively, the standard way as extensively exercised by Hobza and co-workers is to simply
use:
(CCSD(T);CBS) (Y ) (CCSD(T);X) (MP2;) (MP2;X)

Etotal ESCF + Ecorr + Ecorr Ecorr (8.4)
The appropriate keyword is:
!RHF ExtrapolateEP2(2/3,ANO,MP2) TightSCF Conv Bohrs

* int 0 1
O 0 0 0 0 0 0
H 1 0 0 1.81975 0 0
H 1 2 0 1.81975 105.237 0
*
This creates the following output:
Alpha : 5.410 (SCF Extrapolation)

Beta : 2.430 (correlation extrapolation)
SCF energy with basis ano-pVDZ: -76.059178452

SCF energy with basis ano-pVTZ: -76.064774379
Extrapolated CBS SCF energy : -76.065995735 (-0.001221356)
MP2 energy with basis ano-pVDZ: -0.219202872

MP2 energy with basis ano-pVTZ: -0.267058629
Extrapolated CBS correlation energy : -0.295568596 (-0.028509967)
CCSD(T) correlation energy with basis ano-pVDZ: -0.229478341

CCSD(T) - MP2 energy with basis ano-pVDZ: -0.010275470
Estimated CBS total energy : -76.371839801
The estimated correlation energy is not really bad within 3 mEh from the basis set limit.
Using the ExtrapolateEP2(n/m,bas,[method, method-details]) keyword one can use a generalization

of the above method where instead of MP2 any available correlation method can be used as described in
Ref. [83]. method is optional and can be either MP2 or DLPNO-CCSD(T), the latter being the default. In
case the method is DLPNO-CCSD(T) in the method-details option one can ask for LoosePNO, NormalPNO
or TightPNO.
(CCSD(T);CBS) (CCSD(T);X) (M;CBS) (M;X)

Ecorr Ecorr + Ecorr (X, X + 1) Ecorr (8.5)
Here M represents any correlation method one would like to use. For the previous water molecule the input
of a calculation that uses DLPNO-CCSD(T (that is the default now) instead of MP2 would look like:
! RHF ExtrapolateEP2(2/3,cc,DLPNO-CCSD(T)) TightSCF Conv Bohrs

* int 0 1
O 0 0 0 0 0 0
H 1 0 0 1.81975 0 0
H 1 2 0 1.81975 105.237 0
*
and it would produce the following output:



CCSD(T) correlation energy with basis cc-pVDZ: -0.214699894

CCSD(T) - MDCI energy with basis cc-pVDZ: -0.000117258
which is less than 2 mEh from the basis set limit. Finally it was shown [83] that instead of extrapolating
the cheap method, M, using cardinal numbers X and X + 1 it is better to use cardinal numbers X + 1 and
X + 2.
(CCSD(T);CBS) (CCSD(T);X) (M;CBS) (M;X)

Ecorr Ecorr + Ecorr (X + 1, X + 2) Ecorr (8.6)
This can be done using the ExtrapolateEP3(bas,[method,method-details]) keyword:
! RHF ExtrapolateEP3(CC) TightSCF Conv Bohrs
and the corresponding output would be:


SCF energy with basis cc-pVQZ: -76.064381269

MDCI energy with basis cc-pVQZ: -0.295229881
CCSD(T) correlation energy with basis cc-pVDZ: -0.214699894

CCSD(T) - MDCI energy with basis cc-pVDZ: -0.000117257
For the ExtrapolateEP2, and ExtrapolateEP3 keywords the default cheap method is the DLPNO-CCSD(T)
with the NormalPNO thresholds. There also available options with MP2, and DLPNO-CCSD(T) with
LoosePNO and TightPNO settings.
8.1.3.6 Explicitly Correlated MP2 and CCSD(T) Calculations
A physically perhaps somewhat more satisfying alternative to basis set extrapolation is the theory of explicit
correlation. In this method terms are added to the wavefunction Ansatz that contain the interelectronic
coordinates explicitly (hence the name explicit correlation). Initially these terms were linear in the
interelectronic distances (R12-methods). However, it has later been found that better results can be
obtained by using other functions, such as an exponential, of the the interelectronic distance (F12-methods).
These methods are known to yield near basis set limit results for correlation energies in conjunction with
much smaller orbital basis sets.
In applying these methods several points are important:
Special orbital basis sets are at least advantageous. The development of such basis sets is still in its
infancy. For a restricted range of elements the basis sets cc-pVnZ-F12 are available (where n = D, T,
Q) and are recommended. Note, that other than their names suggest, these are a fair bit larger than
regular double, triple or quadruple-zeta basis sets
In addition to an orbital basis set, a near-complete auxiliary basis set must be specified. This is the
so-called CABS basis. For the three basis sets mentioned above these are called cc-pVnZ-F12-CABS.
If you have elements that are not covered you are on your own to supply a CABS basis set. CABS
basis sets can be read into ORCA in a way analogous to RI auxiliary basis sets (replace AUX by
CABS in the input
if the RI approximation is used in conjunction with F12, a third basis set is required - this can be the
regular auxiliary /C basis, but we recommend to step one level up in the auxiliary basis set (e.g. use
a cc-pVTZ/C fitting basis in conjunction with cc-pVDZ-F12)
It is perfectly feasible to use RIJCOSX or RI-JK at the same time. In this case, you should provide a
fourth basis set for the Coulomb fitting
RHF and UHF are available, ROHF not.
Gradients are not available

Doing explicitly correlated MP2 calculations is straightforward. For example look at the following calculation
on the water molecule at a given geometry:
#
! F12-MP2 cc-pVDZ-F12 cc-pVDZ-F12-CABS VeryTightSCF PModel
* xyz 0 1
O 0.000000000000 0.000000000000 0.369372944000
H 0.783975899000 0.000000000000 -0.184686472000
H -0.783975899000 0.000000000000 -0.184686472000
*
and similary in conjunction with the RI approximation:
#
! F12-RI-MP2 cc-pVDZ-F12 cc-pVDZ-F12-CABS cc-pVTZ/C VeryTightSCF PModel
* xyz 0 1
O 0.000000000000 0.000000000000 0.369372944000
H 0.783975899000 0.000000000000 -0.184686472000
H -0.783975899000 0.000000000000 -0.184686472000
*
The output is relatively easy to interprete:
-----------------
RI-MP2-F12 ENERGY
-----------------
EMP2 correlation Energy : -0.286878725335

F12 correction : -0.013254851100
-----------------
MP2 basis set limit estimate : -0.300133576436
Hartree-Fock energy : -76.066649848665

(2)_CABS correction to EHF : -0.000124410174
-----------------
HF basis set limit estimate : -76.066774258839
MP2 total energy before F12 : -76.353528574000

Total F12 correction : -0.013379261275
-----------------
Final basis set limit MP2 estimate : -76.366907835275
It consists of several parts. The first is the regular (RI-)MP2 correlation energy in the orbitals basis followed
by the additive MP2 correction that are combined to provide a MP2 correlation energy basis set limit estimate.
The second part consists of an estimate in the error in the underlying SCF energy. This is the (2) CABS
correction. The combination of the SCF energy with this correction yields an estimate of the SCF basis set
limit. The correction will typically undershoot somewhat, but the error is very smooth. Finally, the corrected
correlation energy and the corrected SCF energy are added to yield the F12 total energy estimate at the
basis set limit.
Lets look at some results and compare to extrapolation:
#
# Correlation energies of the water molecule: extrapolation versus F12
#
# cc-pVDZ MP2: -0.201 380 894
# T : -0.261 263 141
# Q : -0.282 661 311
# T/Q : -0.298 276 192
# Q/5 : -0.300 598 282
# F12-DZ : -0.295 775 804
# RI-F12-DZ : -0.295 933 560 (cc-pVDZ/C)
# -0.295 774 489 (cc-pVTZ/C)
# F12-TZ : -0.299 164 006
# RI-F12-TZ : -0.299 163 478 (cc-pVQZ/C)
# F12-QZ : -0.300 130 086
It is obvious that extrapolated and F12 correlation energies converge to the same number (in this case around
300 mEh). The best extrapolated result is still below the F12 result (this would primarily be meaningful in
a variational calculation). However, first of all this was an expensive extrapolation and second, the small
residual F12 error is very smooth and cancels in energy differences. In any case, already the F12-double-zeta
(where double zeta is to be interpreted rather loosely) brings one into within 5 mEh of the basis set limit
correlation energy and the F12-triple-zeta calculation to within 1 mEh, which is impressive.
The additional effort for the F12 calculation is rather high, since five types of additional two-electron integrals
need to be calculated. Both, integrals in CABS space and in the original orbital (OBS) space must be
calculated and mixed Fock matrices are also required. Hence, one may wonder, whether a double-zeta F12
calculation actually saves any time over, say, a quadruple-zeta regular calculation. The actual answer to this
question is: NO. Given all possibilities of obtained approximate MP2 and SCF energies, we have investigated
the question of how to obtain MP2 basis set limit energies most efficiently in some detail. The results show
that in terms of timings, basis set extrapolation in combination with RI-JK is the method of choice in
combination for MP2. [84] However, energy differences are more reliable with F12-MP2. In combination with
RI-JK or RIJCOSX F12-MP2 becomes also competitive in terms of computational efficiency.
This situation is different in the case of coupled-cluster methods, where F12 methods outperform extrapolation
and are the method of choice.
For coupled-cluster theory, everything works in a very similar fashion:
# the keywords
! F12-CCSD(T)
# and
! CCSD(T)-F12
# are equivalent
A special feature of ORCA that can use large amounts of time, is to use the RI approximation only for the
F12-part. The keyword here is:
! F12/RI-CCSD(T)
# or
! CCSD(T)-F12/RI
Everything else works as described for F12-MP2.
8.1.3.7 Frozen Core Options
In Coupled Cluster calculations the Frozen Core (FC) approximation is applied by default. This implies that
the core electrons are not included in the correlation treatment, since the inclusion of dynamic correlation in
the core electrons usually effects relative energies insignificantly.
The frozen core option can be switched on or off with ! FrozenCore or ! NoFrozenCore in the simple
input. More information and further options are given in section 9.8 and in section 9.10.4.1.
8.1.3.8 Local Coupled Pair and Coupled-Cluster Calculations
ORCA features a special set of local correlation methods. The prevalent local coupled-cluster approaches
date back to ideas of Pulay and have been extensively developed by Werner, Sch utz and co-workers. They
use the concept of correlation domains in order to achieve linear scaling with respect to CPU, disk and main
memory. While the central concept of electron pairs is very similar in both approaches, the local correlation
methods in ORCA follow a completely different and original philosophy.
In ORCA rather than trying to use sparsity, we exploit data compression. To this end two concepts are
used: (a) localization of internal orbitals, which reduces the number of electron pairs to be correlated since
the pair correlation energies are known to fall off sharply with distance; (b) use of a truncated pair specific
natural orbital basis to span the significant part of the virtual space for each electron pair. This guarantees
the fastest convergence of the pair wavefunction and a nearly optimal convergence of the pair correlation
energy while not introducing any real space cut-offs or geometrically defined domains. These PNOs have
been used previously by the pioneers of correlation theory. However, as discussed in the original papers, the
way in which they have been implemented into ORCA is very different. For a full description of technical
details and numerical tests see:
F. Neese, A. Hansen, D. G. Liakos: Efficient and accurate local approximations to the coupled-cluster
singles and doubles method using a truncated pair natural orbital basis. [82]
F. Neese, A. Hansen, F. Wennmohs, S. Grimme: Accurate Theoretical Chemistry with Coupled

Electron Pair Models. [85]
F. Neese, F. Wennmohs, A. Hansen:Efficient and accurate local approximations to coupled electron

pair approaches. An attempt to revive the pair-natural orbital method. [86]
D. G. Liakos, A. Hansen, F. Neese: Weak molecular interactions studied with parallel implementations
of the local pair natural orbital coupled pair and coupled-cluster methods. [87]
A. Hansen, D. G. Liakos, F. Neese: Efficient and accurate local single reference correlation methods for
high-spin open-shell molecules using pair natural orbitals. [88]
C. Riplinger, F. Neese: An efficient and near linear scaling pair natural orbital based local coupled-cluster
method. [89]
C. Riplinger, B. Sandhoefer, A. Hansen, F. Neese: Natural triple excitations in local coupled-cluster

calculations with pair natural orbitals. [90]
C. Riplinger, P. Pinski, U. Becker, E. F. Valeev, F. Neese: Sparse maps - A systematic infrastructure

for reduced-scaling electronic structure methods. II. Linear scaling domain based pair natural orbital
coupled cluster theory. [91]
D. Datta, S. Kossmann, F. Neese: Analytic energy derivatives for the calculation of the first-order
molecular properties using the domain-based local pair-natural orbital coupled-cluster theory [92]
M. Saitow, U. Becker, C. Riplinger, E. F. Valeev, F. Neese: A new linear scaling, efficient and accurate,
open-shell domain based pair natural orbital coupled cluster singles and doubles theory. [93]
In 2013, the so-called DLPNO-CCSD method (domain based local pair natural orbital) was introduced. [89]
This method is near linear scaling with system size and allows for giant calculations to be performed. In 2016,
significant changes to the algorithm were implemented leading to linear scaling with system size concerning
computing time, hard disk and memory consumption. [91] The principle idea behind DLPNO is the following:
it became clear early on that the PNO space for a given electron pair (ij) is local and located in the same
region of space as the electron pair (ij). In LPNO-CCSD this locality was partially used in the local fitting
to the PNOs (controlled by the parameter TCutMKN). However, the PNOs were expanded in canonical
virtual orbitals which led to some higher order scaling steps. In DLPNO, the PNOs are expanded in the set
of projected atomic orbitals:
X
|
i = 1 |ii hi| |i (8.7)
i
where |i is an atomic orbital and |ii refers to an occupied molecular orbital. Such projected orbitals are an
overcomplete representation of the virtual space. The projected orbital | i is located in the same region of
space as |i and hence can be assigned to atomic centers. This has first been invented and used by Pulay and
Saebo [94] in their pioneering work on local correlation methods and widely exploited by Werner, Sch utz and
co-workers in their local correlation approaches. [95, 96] DLPNO-CCSD goes one step further in expanding

the PNOs a ij of a given pair (ij) as:
X ij
a
ij = da |
i (8.8)
{ij}

where {ij} is the domain of atoms (range of

) that is associated with the electron pair ij. The advantage
of the PNO method is, that these domains can be chosen to be large (>15-20 atoms) without compromising
the efficiency of the method.
The comparison between LPNO-CCSD and DLPNO-CCSD is shown in Figure 8.5. It is obvious that
DLPNO-CCSD is (almost) never slower than LPNO-CCSD. However, its true advantages do become most
apparent for molecules with more than approximately 60 atoms. The triples correction, that was added with
our second paper from 2013, shows a perfect linear scaling, as is shown in part (a) of Figure 8.5. For large
systems it adds about 10%20% to the DLPNO-CCSD computation time, hence its addition is possible for
all systems for which the latter can still be obtained. Since 2016, the entire DLPNO-CCSD(T) algorithm is
linear scaling. The improvements of the linear-scaling algorithm, compared to DLPNO2013-CCSD(T), start
to become significant at system sizes of about 300 atoms, as becomes evident in part (b) of Figure 8.5.
(a) DLPNO2013 Scaling (b) DLPNO Scaling
Figure 8.5: a) Scaling behaviour of the canonical CCSD, LPNO-CCSD and DLPNO2013-CCSD(T)
methods. It is obvious that only DLPNO2013-CCSD and DLPNO2013-CCSD(T) can
be applied to large molecules. The advantages of DLPNO2013-CCSD over LPNO-
CCSD do not show before the system has reached a size of about 60 atoms. b) Scaling
behaviour of DLPNO2013-CCSD(T), DLPNO-CCSD(T) and RHF using RIJCOSX. It
is obvious that only DLPNO-CCSD(T) can be applied to truly large molecules, is faster
than the DLPNO2013 version, and even has a crossover with RHF at about 400 atoms.
Using the DLPNO-CCSD(T) approach it was possible for the first time (in 2013) to perform a CCSD(T)
level calculation on an entire protein (Crambin with more than 650 atoms, Figure 8.6). While the calculation
using a double-zeta basis took about 4 weeks on one CPU with DLPNO2013-CCSD(T), it takes only about 4
days to complete with DLPNO-CCSD(T). With DLPNO-CCSD(T) even the triple-zeta basis calculation can
be completed within reasonable time, taking 2 weeks on 4 CPUs.
Figure 8.6: Structure of the Crambin protein - the first protein to be treated with a CCSD(T) level
ab initio method
The use of the LPNO (and DLPNO) methods is simple and requires little special attention from the user:
# Local Pair Natural Orbital Test

! cc-pVTZ cc-pVTZ/C LPNO-CCSD TightSCF
# or
! cc-pVTZ cc-pVTZ/C DLPNO-CCSD TightSCF
%maxcore 2000
# these are the default values - they need not to be touched!

%mdci TCutPNO 3.33e-7 # cutoff for PNO occupation numbers. This
is the main truncation parameter
TCutPairs 1e-4 # cut-off for estimated pair correlation energies.
This exploits the locality in the internal space
TCutMKN 1e-3 # this is a technical parameter here that controls the domain
size for the local fit to the PNOs. It is conservative.
end
* xyz 0 1
... (coordinates)
*
Using the well tested default settings, the LPNO-CEPA (LPNO-CPF, LPNO-VCEPA), LPNO-QCISD and
LPNO-CCSD (LPNO-pCCSD) methods6 can be run in strict analogy to canonical calculations and should
approximate the canonical result very closely. In fact, one should not view the LPNO methods as new model
6
As a technical detail: The closed-shell LPNO QCISD and CCSD come in two technical variants - LPNO1-
CEPA/QCISD/CCSD and LPNO2-CEPA/CCSD/QCISD. The 2 variants consume less disk space but are also
slightly less accurate than the 1 variants. This is discussed in the original paper in the case of QCISD and CCSD.
For the sake of accuracy, the 1 variants are the default. In those cases, where 1 can still be performed, the
chemistry - they are designed to reproduce the canonical results, including BSSE. This is different from the
domain based local correlation methods that do constitute a new model chemistry with properties that are
different from the original methods.
In some situations, it may be appropriate to adapt the accuracy of the calculation. Sensible defaults have
been determined from extensive benchmark calculations and are accessible via LoosePNO, NormalPNO and
TightPNO keywords in the simple input line. [97]
These keywords represent the recommended way to control the accuracy of DLPNO calculations as follows.
Manual changing of thresholds beyond these specifying these keywords is usually discouraged.
# Tight settings for increased accuracy, e.g. when investigating

# weak interactions or conformational equilibria
! cc-pVTZ cc-pVTZ/C DLPNO-CCSD(T) TightPNO TightSCF
# OR: Default settings (no need to give NormalPNO explicitly)

# Useful for general thermochemistry
! cc-pVTZ cc-pVTZ/C DLPNO-CCSD(T) NormalPNO TightSCF
# OR: Loose settings for rapid estimates

! cc-pVTZ cc-pVTZ/C DLPNO-CCSD(T) LoosePNO TightSCF
%maxcore 2000
* xyz 0 1
... (coordinates)
*
Since ORCA 4.0, the linear-scaling DLPNO implementation described in reference [91] is the default DLPNO
algorithm. However, for comparison, the first DLPNO implementation from references [89] and [90] can still
be called by using the DLPNO2013 prefix instead of the DLPNO- prefix.
# DLPNO-CCSD(T) calculation using the 2013 implementation

! cc-pVTZ cc-pVTZ/C DLPNO2013-CCSD(T)
# DLPNO-CCSD(T) calculation using the linear-scaling implementation

! cc-pVTZ cc-pVTZ/C DLPNO-CCSD(T)
* xyz 0 1
... (coordinates)
*
computational efficiency of both approaches is not grossly different. For LPNO CCSD there is also a third variant
(LPNO3-CCSD, also in the open-shell version) which avoids neglecting the dressing of the external exchange
operator. However, the results do not differ significantly from variant 1 but the calculations will become more
expensive. Thus it is not recommend to use variant 3. Variant 2 is not available in the open-shell version.
Since ORCA 4.0, the high-spin open-shell version of the DLPNO-CISD/QCISD/CCSD implementations
have been made available on top of the same machinery as the 2016 version of the RHF-DLPNO-CCSD
code. The present UHF-DLPNO-CCSD is designed to be an heir to the UHF-LPNO-CCSD and serves
as a natural extension to the RHF-DLPNO-CCSD. A striking difference between UHF-LPNO and newly
developed UHF-DLPNO methods is that the UHF-DLPNO approach gives the identical results to that of the
RHF variant when applied to the closed-shell species while the UHF-LPNO does not. Usage of this program
is quite straightforward and shown below:
# (1) In case of ROHF reference

! ROHF DLPNO-CCSD def2-TZVPP def2-TZVPP/C TightSCF TightPNO
# (2) In case of UHF reference, the QROs are constructed first and used for
# the open-shell DLPNO-CCSD computations
! UHF DLPNO-CCSD def2-TZVPP def2-TZVPP/C TightSCF TightPNO
# (3) In case that the UKS are specified, the QROs are constructed first and used as
# "unconverged" UHF orbitals for the open-shell DLPNO-CCSD computations.
! UKS CAM-B3LYP DLPNO-CCSD def2-TZVPP def2-TZVPP/C TightSCF TightPNO
Note that this implementation is dedicated to the closed-shell and high-spin open-shell species. For spin-
polarized systems, the UHF-LPNO-CCSD or Mk-LPNO-CCSD are available in addition to DLPNO-NEVPT2.
The same set of truncation parameters as the closed-shell DLPNO-CCSD is used also in case of open-shell
DLPNO. The open-shell DLPNO-CCSD produces more than 99.9 % of the canonical CCSD correlation energy
as in case of the closed-shell variant. This feature is certainly different from the UHF-LPNO methods because
the open-shell DLPNO-CCSD is re-designed from scratch on the basis of a new PNO ansatz which makes use
of the high-spin open-shell NEVPT framework. The computational timings of the UHF-DLPNO-CCSD and
RIJCOSX-UHF for linear alkane chains in triplet state are shown in Figure 8.7.
103
200
RCOSX-UHF
Computational Time (sec.)
150 UHF-DLPNO-CCSD
100
50
0
40 60 80 100 120 140 160 180
Number of Carbon Atoms
Figure 8.7: Computational times of RIJCOSX-UHF and UHF-DLPNO-CCSD for the linear alkane
chains (Cn Hn+2 ) in triplet state with def2-TZVPP basis and default frozen core settings.
4 CPU cores and 128 GB of memory were used on a single cluster node.
Although those systems are somewhat idealized for the DLPNO method to best perform, it is clear that the
preceding RIJCOSX-UHF is the rate-determining step in the total computational time for large examples. In
the open-shell DLPNO implementations, SOMOs are included not only in the occupied space but also in the
PNO space in the preceding integral transformation step. This means the presence of more SOMOs may lead
to more demanding PNO integral transformation and DLPNO-CCSD iterations. The illustrative examples
include active site model of the [NiFe] Hydrogenase in triplet state and the oxygen evolving complex (OEC)
in the high-spin state, which are shown in Figures 8.8 and 8.9, respectively. With def2-TZVPP basis set and
NormalPNO settings, a single point calculation on [NiFe] Hydrogenase (Figure 8.8) took approximately 45
hours on a single cluster node by using 4 CPU cores of Xeon E5-2670 R
. A single point calculation on the
OEC compound (Figure 8.9) with the same computational settings finished in 44 hours even though the
number of AO in this system is even fewer than the Hydrogenase: the Hydrogenase active site model and
OEC involve 4007 and 2606 AO basis functions, respectively. Special care should be taken if the system
possess more than ten SOMOs since inclusion of more SOMOs may drastically increase the prefactor of the
calculations. In addition, if the SOMOs are distributed over the entire molecular skeleton, each pair domain
may not be truncated at all; in this case speedup attributed to the domain truncation will not be achieved at
all.
Figure 8.8: Ni-Fe active center in the [NiFe] Hydrogenase in its second-coordination sphere. The
whole model system is composed of 180 atoms.
Figure 8.9: A model compound for the OEC in the S2 state of photosystem II which is comprised
of 238 atoms. In its high-spin state, the OEC possesses 13 SOMOs in total.
Calculation of the orbital-unrelaxed density has been implemented for closed-shell DLPNO-CCSD. This
permits analytical computation of first-order properties, such as multipole moments or electric field gradients.
In order to reproduce conventional unrelaxed CCSD properties to a high degree of accuracy, tighter thresholds
may be needed than given by the default settings. Reading of the reference [92] is recommended. Calculation
of the unrelaxed density is requested as usual:
%MDCI Density Unrelaxed End
There are a few things to be noticed about (D)LPNO methods:
The LPNO methods obligatorily make use of the RI approximation. Hence, a correlation fit set must
be provided.
The DLPNO-CCSD(T) method is applicable to closed-shell or high-spin open-shell species.
Besides the closed-shell version which uses a RHF or RKS reference determinant there is an open-shell
version of the LPNO-CCSD for high-spin open-shell molecules (see original paper) using an UHF or
UKS reference determinant build from quasi-restricted orbitals (QROs, see section 7.8.3). Since the
results of the current open-shell version are slightly less accurate than that of the closed-shell version
it is mandatory to specify if you want to use the closed-shell or open-shell version for calculations of
closed-shell systems, i.e. always put the RHF (RKS) or UHF (UKS) keyword in the simple
keyword line. Open-shell systems can be of course only treated by the open-shell version. Do not mix
results of the closed- and open-shell versions of LPNO methods (e.g. if you calculate reaction
energies of a reaction in which both closed- and open-shell molecules take part, you should use the
open-shell version throughout). This is because the open-shell LPNO results for the closed-shell species
certainly differ from those of closed-shell implementations. This drawback of the open-shell LPNO
methods has led to the development of a brand new open-shell DLPNO approach which converges
to the RHF-DLPNO in the closed-shell limit. Importantly, one can mix the results of closed-
and open-shell versions of DLPNO approaches.
The open-shell version of the DLPNO approach uses a different strategy to the LPNO variant to define
the open-shell PNOs. This ensures that, unlike the open-shell LPNO, the PNO space converges to the
closed-shell counterpart in the closed-shell limit. Therefore, in the closed-shell limit, the open-shell
DLPNO gives identical correlation energy to the RHF variant up to at least the third decimal place.
The perturbative triples correction referred to as, (T), is also available for the open-shell species.
When performing a calculation on the open-shell species with either of canonical/LPNO/DLPNO

methods on top of the Slater determinant constructed from the QROs, a special attention should be
paid on the orbitals energies of those QROs. In some cases, the orbitals energy of the highest SOMO
appear to be higher than that of the lowest VMO. Similarly to this, the orbital energy of the highest
DOMO may appear to higher than that of the lowest SOMOs. In such cases, the CEPA/QCISD/CCSD
iteration may show difficulty in convergence. In the worst case, it just diverges. Most likely, in such
cases, one has to suspect the charge and multiplicity might be wrong. If they are correct, you may
need much prettier starting orbitals and a bit of good luck!
F12-variants will become available in the foreseeable future.
Parallelization is done.
There are three thresholds that can be user controlled that can all be adjusted in the %mdci block: (a)
TCutPNO controls the number of PNOs per electron pair. This is the most critical parameter and has a
default value of 3.33 107 . (b) TCutPairs controls a perturbative selection of significant pairs and has a
default value of 104 . (c) TCutMKN is a technical parameter and controls the size of the fit set for each
electron pair. It has a default value of 103 . All of these default values are conservative. Hence, no
adjustment of these parameters is necessary. All DLPNO-CCSD truncations are bound to these three
truncation parameters and should not almost be touched (Hence they are also not documented :-)).
The preferred way to adjust accuracy when needed is to use the LoosePNO/NormalPNO/TightPNO
keywords. In addition, TightPNO triggers the full iterative (DLPNO-MP2) treatment in the MP2
guess, whereas the other options use a semicanonical MP2 calculation. Tables 8.6 and 8.7 contain the
thresholds used by the current (2016) and old (2013) implementations, respectively.
LPNO-VCEPA/n (n=1,2,3) methods are only available in the open-shell version yet.
LPNO variants of the parameterized coupled-cluster methods (pCCSD, see section 7.8.1) are also
available (e.g. LPNO-pCCSD/1a and LPNO-pC CSD/2a).
The LPNO methods reproduce the canonical energy differences typically better than 1 kcal/mol. This
accuracy exists over large parts of the potential energy surface. Tightening TCutPairs to 1e-5 gives
more accurate results but also leads to significantly longer computation times.
Potential energy surfaces are virtually but not perfectly smooth (like any method that involves cut-offs).
Numerical gradient calculations have been attempted and reported to have been successful.
The LPNO methods do work together with RIJCOSX, RI-JK and also with ANO basis sets and basis
set extrapolation. They also work for conventional integral handling.
The methods behave excellently with large basis sets. Thus, they stay efficient even when large basis
sets are used that are necessary to obtain accurate results with wavefunction based ab initio methods.
This is a prerequisite for efficient computational chemistry applications.
For LPNO-CCSD, calculations with about 1000 basis functions are routine, calculations with about
1500 basis functions are possible and calculations with 2000-2500 basis functions are the limit on
powerful computers. For DLPNO-CCSD much larger calculations are possible. There is virtually no
crossover and DLPNO-CCSD is essentially always more efficient than LPNO-CCSD. Starting from
about 50 atoms the differences become large. The largest DLPNO-CCSD calculation to date featured
>1000 atoms and more than 20000 basis functions!
Using large main memory is not mandatory but advantageous since it speeds up the initial integral
transformation significantly (controlled by MaxCore in the %mdci block, see section 7.8.4).
The open-shell versions are about twice as expensive as the corresponding closed-shell versions.
Analytic gradients are not available.
An unrelaxed density implementation is available for closed-shell DLPNO-CCSD, permitting calculation

of first-order properties.
Table 8.6: Accuracy settings for DLPNO coupled cluster (current version).
Setting TCutPairs TCutDO TCutPNO TCutMKN MP2 pair treatment

LoosePNO 103 2 102 1.00 106 103 semicanonical
NormalPNO 104 1 102 3.33 107 103 semicanonical
TightPNO 105 5 103 1.00 107 103 full iterative
Table 8.7: Accuracy settings for DLPNO coupled cluster (deprecated 2013 version).
Setting TCutPairs TCutPNO TCutMKN MP2 pair treatment

LoosePNO 103 1.00 106 103 semicanonical
NormalPNO 104 3.33 107 103 semicanonical
TightPNO 105 1.00 107 104 full iterative
As an example, see the following isomerization reaction that appears to be particularly difficult for DFT:
Isomerizes to:
The results of the calculations (closed-shell versions) with the def2-TZVP basis set (about 240 basis functions)
are shown below:
Method Energy Difference (kcal/mol) Time (min)

CCSD(T) -14.6 92.4
CCSD -18.0 55.3
LPNO-CCSD -18.6 20.0
CEPA/1 -12.4 42.2
LPNO-CEPA/1 -13.5 13.4
The calculations are typical in the sense that: (a) the LPNO methods provide answers that are within 1
kcal/mol of the canonical results, (b) CEPA approximates CCSD(T) more closely than CCSD. The speedups
of a factor of 2 5 are moderate in this case. However, this is also a fairly small calculation. For larger
systems, speedups of the LPNO methods compared to their canonical counterparts are on the order of a
factor >1001000.
8.1.3.9 Cluster in molecules (CIM)
Cluster in molecules (CIM) approach is a linear scaling local correlation method developed by Li and the
coworkers in 2002. [98] It was further improved by Li, Piecuch, Kallay and other groups recently. [99102]
CIM is inspired by the early local correlation method developed by Forner and coworkers. [103] The total
correlation energy of a close-shell molecule can be considered as the summation of correlation energies for
each occupied LMOs.
occ occ
X X 1X ij
Ecorr = Ei = hij||abiTab (8.9)
i i
4
j,ab
For each occupied LMO, it only correlated with its nearby occupied LMOs and virtual MOs. To reproduce
the correlation energies for each occupied LMO, only a subset of occupied and virtual LMOs are needed in
the correlation calculation. Instead of doing the correlation calculation of the whole molecule, the correlation
energy can be obtained by calculating various subsystems. The CIM approach implemented in ORCA is
almost following the algorithm proposed by Guo and coworkers. [102] However, to avoid the real space cutoff,
we use the differential overlap integral (DOI) instead of distance threshold. There is only one parameter
CIMTHRESH in CIM approach, which controls the numbers of MOs in subsystems. If the DOI between
LMO i and LMO j is larger than CIMTHRESH, j will be included into the MO domain of i. By including
all nearby LMO of i, one can construct a subsystem for MO i. The default value of CIMTHRESH is
0.001, which can reproduce about 99.8% of the correlation energies. If accurate results are needed, the
tight CIMTHRESH must be used. To reproduce the correlation energies at different levels, CIM can invoke
different correlation methods to calculate the subsystems. In ORCA the CIM-RI-MP2, CIM-CCSD(T)
and CIM-DLPNO-CCSD(T) is available now, which can reproduce the energies of RI-MP2, CCSD(T) and
DLPNO-CCSD(T).
The usage of CIM in ORCA is simple. For CIM-RI-MP2,
#
# CIM-RI-MP2 calculation
#
! RI-MP2 cc-pVDZ cc-pVDZ/C CIM
%CIM
CIMTHRESH 0.0005 # Default value is 0.001
end
* xyzfile 0 1 CIM.xyz
For CIM-DLPNO-CCSD(T),
#
# CIM-DLPNO-CCSD calculation
#
! DLPNO-CCSD(T) cc-pVDZ cc-pVDZ/C CIM
* xyzfile 0 1 CIM.xyz
The parallelization of CIM have been implemented. However, it is only parallelized within one subsystem
calculation. The parallelization between subsystems will be released in the next generation of ORCA . The
generalization of CIM from close-shell to open-shell will also be implemented in near furture.
8.1.3.10 Arbitrary Order Coupled-Cluster Calculations
ORCA features an interface to Kallays powerful MRCC program. This program must be obtained separately.
The interface is restricted to single point energies but can be used for rigid scan calculations or numerical
frequencies.
The use of the interface is simple:
#
# Test the MRCC code of Mihael Kallay
#
! cc-pVDZ Conv SCFConv10 UseSym
%mrcc method "CCSDT"

ETol 9
end
* xyz 0 1
F 0 0 0
H 0 0 0.95
*
The Method string can be any of:
# The excitation level specification can be anything

# like SD, SDT, SDTQ, SDTQP etc.
%mrcc method "CCSDT"
"CCSD(T)"
"CCSD[T]"
"CCSD(T)_L" (the lambda version)
"CC3"
"CCSDT-1a"
"CCSDT-1b"
"CISDT"
It is not a good idea, of course, to use this code for CCSD or CCSD(T) or CISD. Its real power lies
in performing the higher order calculations. Open-shell calculations can presently not be done with the
interface.
8.1.4 Density Functional Theory
8.1.4.1 Standard Density Functional Calculations
Density functional calculations are as simple to run as HF calculations. In this case you may want to adjust
the integration grid and you almost certainly will want to use the RI-J approximation in case that you have
a LDA, GGA or meta-GGA functional (non-hybrid functional). For hybrid functionals, the RIJCOSX and
RI-JK approximations offer large speedups.
For example, consider this B3LYP calculation on the cyclohexane molecule.
# Test a simple DFT calculation

! RKS B3LYP SVP Grid4
* xyz 0 1
C -0.79263 0.55338 -1.58694
C 0.68078 0.13314 -1.72622
C 1.50034 0.61020 -0.52199
C 1.01517 -0.06749 0.77103
C -0.49095 -0.38008 0.74228
C -1.24341 0.64080 -0.11866
H 1.10490 0.53546 -2.67754
H 0.76075 -0.97866 -1.78666
H -0.95741 1.54560 -2.07170
H -1.42795 -0.17916 -2.14055
H -2.34640 0.48232 -0.04725
H -1.04144 1.66089 0.28731
H -0.66608 -1.39636 0.31480
H -0.89815 -0.39708 1.78184
H 1.25353 0.59796 1.63523
H 1.57519 -1.01856 0.93954
H 2.58691 0.40499 -0.67666
H 1.39420 1.71843 -0.44053
*
If you want an accurate single point energy then it is wise to choose TightSCF, select a basis set of at
least valence triple-zeta plus polarization quality (e.g. def2-TZVP) and also to move one step up in the DFT
integration grid (i.e. Grid4).
8.1.4.2 DFT Calculations with RI
DFT calculations that do not require the HF exchange to be calculated (non-hybrid DFT) can be very
efficiently executed with the RI-J approximation. It leads to very large speedups at essentially no loss of
accuracy. The use of the RI-J approximation may be illustrated for a medium sized organic molecule -
Penicillin:
# RI-DFT calculation on the Penicillin molecule

! RKS BP86 RI SVP def2/J TightSCF
* xyz 0 1
N 3.17265 1.15815 -0.09175
C 2.66167 0.72032 1.18601
C 4.31931 0.59242 -0.73003
C 2.02252 1.86922 -0.54680
C 1.37143 1.52404 0.79659
S 2.72625 -1.05563 0.80065
C 4.01305 -0.91195 -0.52441
C 5.58297 1.09423 -0.06535
O 1.80801 2.36292 -1.62137
N 0.15715 0.73759 0.70095
C 5.25122 -1.72918 -0.12001
C 3.41769 -1.50152 -1.81857
O 6.60623 1.14077 -0.91855
O 5.72538 1.40990 1.08931
C -1.08932 1.35001 0.75816
C -2.30230 0.45820 0.54941
O -1.19855 2.53493 0.96288
O -3.48875 1.21403 0.57063
C -4.66939 0.59150 0.27339
C -4.84065 -0.79240 0.11956
C -5.79523 1.39165 0.03916
C -6.07568 -1.34753 -0.22401
C -7.03670 0.85454 -0.30482
C -7.18253 -0.52580 -0.43612
H 3.24354 1.09074 2.02120
H 4.33865 0.87909 -1.77554
H 1.26605 2.42501 1.39138
H 0.17381 -0.25857 0.47675
H 6.05024 -1.64196 -0.89101
H 5.67754 -1.39089 0.85176
H 5.01118 -2.81229 -0.01401
H 2.50304 -0.95210 -2.14173
H 4.15186 -1.44541 -2.65467
H 3.14138 -2.57427 -1.69700
H 7.29069 1.46408 -0.31004
H -2.21049 -0.02915 -0.44909
H -2.34192 -0.28647 1.37775
H -4.00164 -1.48999 0.26950
H -5.69703 2.48656 0.12872
H -6.17811 -2.44045 -0.33185
H -7.89945 1.51981 -0.47737
H -8.15811 -0.96111 -0.71027
*
The job has 42 atoms and 430 contracted basis functions. Yet, it executes in just a few minutes elapsed time
on any reasonable personal computer.
NOTES:
The RI-J approximation requires an auxiliary basis set in addition to a normal orbital basis set. For
the Karlsruhe basis sets there is the universal auxiliary basis set of Weigend that is called with the
name def2/J (all-electron up to Kr). When scalar relativistic Hamiltonians are used (DKH or ZORA)
along with all-electron basis sets, then a general-purpose auxiliary basis set is the SARC/J that covers
most of the periodic table. Other choices are documented in sections 6.3 and 9.3.
For pure functionals the use of RI-J with the def2/J auxiliary basis set is the default.
Since DFT is frequently applied to open-shell transition metals we also show one (more or less trivial) example
of a Cu(II) complex treated with DFT.
! UKS BP86 RI SV def2/J SlowConv

%base "temp"
* xyz -2 2
Cu 0 0 0
Cl 2.25 0 0
Cl -2.25 0 0
Cl 0 2.25 0
Cl 0 -2.25 0
*
$new_job
! UKS B3LYP NoRI TZVP TightSCF MORead

%moinp "temp.gbw"
%scf GuessMode CMatrix
end
* xyz -2 2
Cu 0 0 0
Cl 2.25 0 0
Cl -2.25 0 0
Cl 0 2.25 0
Cl 0 -2.25 0
*
Although it would not have been necessary for this example, it shows a possible strategy how to converge such
calculations. First a less accurate but fast job is performed using the RI approximation, a GGA functional
and a small basis set without polarization functions. Note that a larger damping factor has been used in order
to guide the calculation (SlowConv). The second job takes the orbitals of the first as input and performs a
more accurate hybrid DFT calculation. A subtle point in this calculation on a dianion in the gas phase is the
command GuessMode CMatrix that causes the corresponding orbital transformation to be used in order to
match the orbitals of the small and the large basis set calculation. This is always required when the orbital
energies of the small basis set calculation are positive as will be the case for anions.
8.1.4.3 HartreeFock and Hybrid DFT Calculations with RIJCOSX
Frustrated by the large difference in execution times between pure and hybrid functionals, we have been
motivated to study approximations to the Hartree-Fock exchange term. The method that we have finally
come up with is called the chain of spheres COSX approximation and may be thought of as a variant
of the pseudo-spectral philosophy. Essentially, in performing two electron integrals, the first integration is
done numerically on a grid and the second (involving the Coulomb singularity) is done analytically.7 Upon
combining this treatment with the Split-RI-J method for the Coulomb term (thus, you do need a Coulomb
fitting basis!), we have designed the RIJCOSX approximation that can be used to accelerate Hartree-Fock
and hybrid DFT calculations. Note that this introduces another grid on top of the DFT integration grid
which is usually significantly smaller.
In particular for large and accurate basis sets, the speedups obtained in this way are very large - we have
observed up to a factor of sixty! The procedure is essentially linear scaling such that large and accurate
calculations become possible with high efficiency. The RIJCOSX approximation is basically available
throughout the program. The errors are on the order of 1 kcal mol1 or less in the total energies as well as in
energy differences and can be made smaller with larger than the default grids or by running the final SCF
cycle without this approximation. The impact on bond distances is a fraction of a pm, angles are better than
a few tenth of a degree and soft dihedral angles are good to about 1 degree. To the limited extent to which it
has been tested, vibrational frequencies are roughly good to 210 wavenumbers with the default settings.
The use of RIJCOSX is very simple:
! B3LYP def2-TZVPP def2/J TightSCF RIJCOSX

...
One thing to be mentioned in correlation calculations with RIJCOSX is that the requirements for the SCF
and correlation fitting bases are quite different. We therefore now support two different auxiliary basis sets in
the same run:
! RI-MP2 def2-TZVPP def2/J def2-TZVPP/C TightSCF RIJCOSX

...
CAUTION:
This feature does NOT work in multiple job inputs

7
For algorithmic and theoretical details see: [104].
8.1.4.4 HartreeFock and Hybrid DFT Calculations with RI-JK
An alternative algorithm for accelerating the HF exchange in hybrid DFT or HF calculations is to use the RI
approximation for both Coulomb and exchange. This is implemented in ORCA for SCF single point energies
but not for gradients.
! RHF def2-TZVPP def2/JK RI-JK

...
The speedups for small molecules are better than for RIJCOSX, for medium sized molecules (e.g. (gly)4 )
similar, and for larger molecules RI-JK is less efficient than RIJCOSX. The errors of RI-JK are usually below
1 mEh and the error is very smooth (smoother than for RIJCOSX). Hence, for small calculations with large
basis sets, RI-JK is a good idea, for large calculations on large molecules RIJCOSX is better.
NOTES:
For RI-JK you will need a larger auxiliary basis set. For the Karlsruhe basis set, the universal def2/JK
and def2/JKsmall basis sets are available. They are large and accurate.
For UHF RI-JK is roughly twice as expensive as for RHF. This is not true for RIJCOSX.
RI-JK is available for conventional and direct runs and also for ANO bases. There the conventional
mode is recommended.
A comparison of the RIJCOSX and RI-JK methods (taken from Ref. [105]) for the (gly)2 , (gly)4 and (gly)8 is
shown below (wall clock times in second for performing the entire SCF):
Def2-SVP Def2-TZVP(-df ) Def2-TZVPP Def2-QZVPP

(gly)2 Default 105 319 2574 27856
RI-JK 44 71 326 3072
RIJCOSX 70 122 527 3659
(gly)4 Default 609 1917 13965 161047
RI-JK 333 678 2746 30398
RIJCOSX 281 569 2414 15383
(gly)8 Default 3317 12505 82774
RI-JK 3431 5452 16586 117795
RIJCOSX 1156 2219 8558 56505
It is obvious from the data that for small molecules the RI-JK approximation is the most efficient choice. For
(gly)4 this is already no longer obvious. For up to the def2-TZVPP basis set, RI-JK and RIJCOSX are almost
identical and for def2-QZVPP RIJCOSX is already a factor of two faster than RI-JK. For large molecules like
(gly)8 with small basis sets RI-JK is not a big improvement but for large basis set it still beats the normal
4-index calculation. RIJCOSX on the other hand is consistently faster. It leads to speedups of around 10 for
def2-TZVPP and up to 50-60 for def2-QZVPP. Here it outperforms RI-JK by, again, a factor of two.
8.1.4.5 DFT Calculations with Second Order Perturbative Correction (Double-Hybrid

Functionals)
There is a family of functionals which came up in 2006 and were proposed by Grimme [106]. They consist
of a semi-empirical mixture of DFT components and the MP2 correlation energy calculated with the DFT
orbitals and their energies. Grimme referred to his functional as B2PLYP (B88 exchange, 2 parameters that
were fitted and perturbative mixture of MP2 and LYP) a version with improved performance (in particular
for weak interactions) is mPW2PLYP [107] and is also implemented. From the extensive calibration work,
the new functionals appear to give better energetics and a narrower error distribution than B3LYP. Thus, the
additional cost of the calculation of the MP2 energy may be well invested (and is quite limited in conjunction
with density fitting in the RI part). Martin has reported reparameterizations of B2PLYP (B2GP-PLYP,
B2K-PLYP and B2T-PLYP) that are optimized for general-purpose, kinetic and thermochemistry
applications. A fairly new variant of these double-hybrids is the PWPB95 functional by Grimme. For more
information, see the detailed input section.
In this version of ORCA, the method is available for single points, geometry optimizations [108], dipole
moments and other first order properties, as well as for numerical polarizabilities and frequencies.
8.1.4.6 DFT Calculations with Atom-pairwise Dispersion Correction
It is well known that DFT does not include dispersion forces. It is possible to use a simple atom-pairwise
correction to account for the major parts of this contribution to the energy [1, 2, 109]. We have adopted the
code and method developed by Stefan Grimme in this ORCA version. The method is parameterized for many
established functionals (e.g. BLYP, BP86, PBE, TPSS, B3LYP, B2PLYP).8 The atom-pairwise dispersion
correction with Becke-Johnson damping (keyword ! D3BJ) is the default option and will automatically be
invoked by the simple keyword ! D3.
! BLYP D3 def2-QZVPP RI def2/J Opt Grid4
%paras R= 2.5,4.0,16
end
%geom Constraints
{ C 0 C }
{ C 1 C }
end
end
* xyz 0 1
Ar 0.0000000 0.0000000 {R}
H 0.0000000 0.0000000 0.0000000
8
For expert users: The keyword D2, D3ZERO and D3BJ select the empirical 2006 and the atom-pairwise 2010 model,
respectively, with either zero-damping or Becke-Johnson damping. The default is the most accurate D3BJ model.
The outdated model from 2004 [110] is no longer supported and can only be invoked by setting DFTDOPT = 1. The
C6-scaling coefficient can be user defined using e.g. %method DFTDScaleC6 1.2 end
C 0.0000000 0.0000000 -1.0951073

H 0.5163499 0.8943443 -1.4604101
H 0.5163499 -0.8943443 -1.4604101
H -1.0326998 0.0000000 -1.4604101
*
In this example, a BLYP calculation without dispersion correction will show a repulsive potential between
the argon atom and the methane molecule. Using the D3 dispersion correction as shown above, the potential
curve shows a minimum at about 3.13.2 A. The atom-pairwise correction is quite successful and Grimmes
work suggests that this is more generally true. For many systems like stacked DNA basepairs, hydrogen bond
complexes and other weak interactions the atom-pairwise dispersion correction will improve substantially the
results of standard functionals at essentially no extra cost.
8.1.4.7 DFT Calculations with Range-Separated Hybrid Functionals
All range-separated functionals in ORCA use the error function based approach according to Hirao and
coworkers. [111] This allows the definition of DFT functionals that dominate the short-range part by an
adapted exchange functional of LDA, GGA or meta-GGA level and the long-range part by Hartree-Fock
exchange.
CAM-B3LYP, [112] LC-BLYP [113] and members of the B97-family of functionals have been implemented
into ORCA, namely B97, B97X [114] and B97X-D3. [115] Some of them incorporate fixed amounts
of Hartree-Fock exchange (EXX) and/or DFT exchange and they differ in the RS-parameter . In the
case of B97X-D3, the proper D3 correction (employing the zero-damping scheme) should be calculated
automatically. The user is encouraged to check this.
Several restrictions apply to these functionals at the moment. They have only been implemented and tested
for use with the libint integral package and for RHF and UHF single-point, ground state nuclear gradient,
ground state nuclear hessian, TDDFT, and TDDFT nuclear gradient calculations. Only the standard integral
handling (NORI), RIJONX, and RIJCOSX are supported. Do not use these functionals with any other
options.
8.1.5 Quadratic Convergence
Convergence in SCF calculations is not always easy to achieve. One way to go if the calculations converges
only slowly or creeps towards the end of the SCF cycles such that many almost useless cycles are being
performed is to switch on a powerful but somewhat expensive convergence helper the full Newton-Raphson
method. In the neighbourhood of a stable SCF solution this method converges quadratically which means
that after 3-4 Newton-Raphson cycles the calculations are normally converged. However, each cycle consists
of microiterations which are roughly as expensive as one SCF iteration. Thus, each SCF iteration becomes
somewhat expensive but this may be more than compensated by the reduced number of cycles.
Consider the following example of a molecule with a small HOMOLUMO gap ( 1 eV):
! PModel RKS BP86 RI SemiDirect SV(P) def2/J TightSCF
* xyz 0 1
C -2.65720 1.06150 -0.05713
C -1.26906 0.45978 0.20734
C 2.30892 0.84241 -0.90960
C 3.66808 0.62825 -0.22855
C -3.24744 1.63074 1.22706
C 4.55462 1.85858 -0.37334
O -4.34718 2.12974 1.26070
O 4.32817 2.77288 -1.12962
S -0.52934 -0.10251 -1.36344
S 1.32901 -0.69440 -0.81445
N -3.52304 -0.01967 -0.66394
N 3.40865 0.27323 1.21839
O -2.67918 1.59541 2.29262
O 5.54649 1.99821 0.30260
H -2.59274 1.90623 -0.78323
H -3.61570 -0.84473 -0.06839
H -3.16141 -0.34892 -1.56042
H -4.47232 0.30853 -0.85012
H 4.27898 0.16311 1.74252
H 2.86502 0.97910 1.71822
H 2.90558 -0.60935 1.32107
H 4.22252 -0.21720 -0.70036
H 2.45549 1.12172 -1.97850
H 1.78000 1.68539 -0.40747
H -0.61420 1.23221 0.67374
H -1.35568 -0.39162 0.92184
*
$new_job
! PModel RKS BP86 RI SemiDirect SV(P) def2/J TightSCF NRSCF
* xyz 0 1
... etc, coordinates repeated
*
The first job converges as follows:
ITER Energy Delta-E Max-DP RMS-DP [F,P] Damp

0 -1441.4978981358 0.000000000000 0.40179217 0.00763633 0.2989113 0.7000
***Turning on DIIS***
1 -1441.7605061369 -0.262608001152 0.88234063 0.01388441 0.0944375 0.0000
2 -1441.4732415586 0.287264578307 0.11142170 0.00283563 0.2353949 0.7000
3 -1441.7133405526 -0.240098993967 0.13296844 0.00300970 0.1969393 0.7000

4 -1441.8756771704 -0.162336617820 0.12684567 0.00251575 0.1502491 0.7000
5 -1441.9617903323 -0.086113161899 0.09818570 0.00195646 0.1031261 0.7000
6 -1442.0049616788 -0.043171346504 0.26817199 0.00537057 0.0669699 0.0000
7 -1442.0598897762 -0.054928097428 0.04412999 0.00080076 0.0423744 0.0000
8 -1442.0665608972 -0.006671121001 0.04285625 0.00064591 0.0175566 0.0000
9 -1442.0686166076 -0.002055710323 0.02519628 0.00048681 0.0045225 0.0000
10 -1442.0690784815 -0.000461873967 0.00652186 0.00017711 0.0024081 0.0000
11 -1442.0691994344 -0.000120952896 0.00318862 0.00009817 0.0013287 0.0000
12 -1442.0692370986 -0.000037664145 0.00289703 0.00006658 0.0010036 0.0000
13 -1442.0692557561 -0.000018657537 0.00872807 0.00026481 0.0006098 0.0000
14 -1442.0692790592 -0.000023303114 0.00136738 0.00001713 0.0019054 0.0000
15 -1442.0692841649 -0.000005105642 0.00194277 0.00004165 0.0007584 0.0000
16 -1442.0692857367 -0.000001571858 0.00043978 0.00001226 0.0002705 0.0000
17 -1442.0692859394 -0.000000202716 0.00028171 0.00000515 0.0001447 0.0000
18 -1442.0692859977 -0.000000058216 0.00014881 0.00000277 0.0000886 0.0000
**** Energy Check signals convergence ****
Thus, the job converges almost monotonically but it does so somewhat slowly towards the end. The second
job, however:

*** Starting incremental Fock matrix formation ***
0 -1441.4978981358 0.000000000000 0.40179217 0.00763633 0.2989113 0.7000
1 -1441.7605061369 -0.262608001152 0.88234063 0.01388441 0.0944375 0.0000
2 -1441.4732415586 0.287264578307 0.11142170 0.00283563 0.2353949 0.7000
3 -1441.7133405526 -0.240098993967 0.13296844 0.00300970 0.1969393 0.7000
4 -1441.8756771704 -0.162336617820 0.12684567 0.00251575 0.1502491 0.7000
5 -1441.9617903323 -0.086113161899 0.09818570 0.00195646 0.1031261 0.7000
6 -1442.0049616788 -0.043171346504 0.26817199 0.00537057 0.0669699 0.0000
7 -1442.0598897762 -0.054928097428 0.04412999 0.00080076 0.0423744 0.0000
8 -1442.0665608972 -0.006671121001 0.04285625 0.00064591 0.0175566 0.0000
*** Initiating the Newton-Raphson procedure ***
*** Shutting down DIIS ***
*** Removing any level shift ***
ITER Energy Delta-E Grad Rot Max-DP RMS-DP
9 -1442.06861661 -0.0020557103 0.002665 0.000000 0.025196 0.000487
CP-SCF ITERATION 0:
CP-SCF ITERATION 1: 0.000033353
10 -1442.06907848 -0.0004618740 0.008069 0.035908 0.017951 0.000570
CP-SCF ITERATION 0:
11 -1442.06928354 -0.0002050609 0.000678 0.004293 0.002640 0.000063
CP-SCF ITERATION 0:
12 -1442.06928602 -0.0000024726 0.000066 0.000109 0.000051 0.000002
<<< The NR Solver signals convergence >>>
Thus, after reaching the threshold for initiating the Newton-Raphson procedure after nine iterations, the job
takes only three more iterations to converge to the correct solution. Since each micro-iteration in the CP-SCF
procedure roughly corresponds to the formation of one Fock-matrix the second job is still somewhat more
expensive. You cannot always expect the Newton-Raphson procedure to converge9 and it will not converge
at all if your SCF solution is not stable (i.e. if the orbital Hessian has negative eigenvalues) still, it is a
powerfull technique to try if other alternatives do not do well. It is important, however, to bring the SCF
into the radius of convergence of the Newton-Raphson procedure for it to be efficient.
The implementation covers closed-shell and spin-unrestricted Hartree-Fock and DFT calculations.
8.1.6 Counterpoise Correction
In calculating weak molecular interactions the nasty subject of the basis set superposition error (BSSE) arises.
It consists of the fact that if one describes a dimer, the basis functions on A help to lower the energy of
fragment B and vice versa. Thus, one obtains an energy that is biased towards the dimer formation due to
basis set effects. Since this is unwanted, the Boys and Bernardi procedure aims to correct for this deficiency
by estimating what the energies of the monomers would be if they had been calculated with the dimer basis
set. This will stabilize the monomers relative to the dimers. The effect can be a quite sizeable fraction of the
interaction energy and should therefore be taken into account. The original Boys and Bernardi formula for
the interaction energy between fragments A and B is:
AB A B
AB AB AB AB

E = EAB (AB) EA (A) EB (B) EA (AB) EA (A) + EB (AB) EB (B) (8.10)
Y
Here EX (Z) is the energy of fragment X calculated at the optimized geometry of fragment Y with the basis
set of fragment Z. Thus, you need to do a total the following series of calculations: (1) optimize the geometry
AB A B
of the dimer and the monomers with some basis set Z. This gives you EAB (AB), EA (A) and EB (B) (2)
delete fragment A (B) from the optimized structure of the dimer and re-run the single point calculation
AB AB
with basis set Z. This gives you EB (B) and EA (A). (3) Now, the final calculation consists of calculating
AB
the energies of A and B at the dimer geometry but with the dimer basis set. This gives you EA (AB) and
AB
EB (AB).
In order to achieve the last step efficiently, a special notation was put into ORCA which allows you to delete
the electrons and nuclear charges that come with certain atoms but retain the assigned basis set. This trick
consists of putting a : after the symbol of the atom. Here is an example of how to run such a calculation of
the water dimer at the MP2 level (with frozen core):
#
# BSSE test
#
# --------------------------------------------
# First the monomer. It is a waste of course
# to run the monomer twice ...
# --------------------------------------------
! RHF MP2 TZVPP VeryTightSCF XYZFile PModel
%id "monomer"
9
For example try %scf nrstart 0.1 end end to turn on the NRSCF after two iterations in the example above it
will run into trouble.
* xyz 0 1
O 7.405639 6.725069 7.710504
H 7.029206 6.234628 8.442160
H 8.247948 6.296600 7.554030
*
$new_job
%id "monomer"
* xyz 0 1
O 7.405639 6.725069 7.710504
H 7.029206 6.234628 8.442160
H 8.247948 6.296600 7.554030
*
# --------------------------------------------
# now the dimer
# --------------------------------------------
$new_job
%id "dimer"
* xyz 0 1
O 7.439917 6.726792 7.762120
O 5.752050 6.489306 5.407671
H 7.025510 6.226170 8.467436
H 8.274883 6.280259 7.609894
H 6.313507 6.644667 6.176902
H 5.522285 7.367132 5.103852
*
# --------------------------------------------
# Now the calculations of the monomer at the
# dimer geometry
# --------------------------------------------
$new_job
%id "monomer_1"
* xyz 0 1
O 7.439917 6.726792 7.762120
H 7.025510 6.226170 8.467436
H 8.274883 6.280259 7.609894
*
$new_job
%id "monomer_1"
* xyz 0 1
O 5.752050 6.489306 5.407671
H 6.313507 6.644667 6.176902
H 5.522285 7.367132 5.103852
*
# --------------------------------------------
# Now the calculation of the monomer at the
# dimer geometry but with the dimer basis set
# --------------------------------------------
$new_job
%id "monomer_2"
* xyz 0 1
O 7.439917 6.726792 7.762120
O : 5.752050 6.489306 5.407671
H 7.025510 6.226170 8.467436
H 8.274883 6.280259 7.609894
H : 6.313507 6.644667 6.176902
H : 5.522285 7.367132 5.103852
*
$new_job
%id "monomer_2"
* xyz 0 1
O : 7.439917 6.726792 7.762120
O 5.752050 6.489306 5.407671
H : 7.025510 6.226170 8.467436
H : 8.274883 6.280259 7.609894
H 6.313507 6.644667 6.176902
H 5.522285 7.367132 5.103852
*
You obtain the energies:
Monomer : -152.647062118 Eh
Dimer : -152.655623625 Eh -5.372 kcal/mol
Monomer at dimer geometry: -152.647006948 Eh 0.035 kcal/mol
Same with AB Basis set : -152.648364970 Eh -0.818 kcal/mol
Thus, the corrected interaction energy is:

-5.372 kcal/mol - (-0.818-0.035)=-4.52 kcal/mol
8.1.7 Complete Active Space Self-Consistent Field Method
8.1.7.1 Introduction
There are several situations where a complete-active space self-consistent field (CASSCF) treatment is a good
idea:
Wavefunctions with significant multireference character arising from several nearly degenerate configu-
rations (static correlation)
Wavefunctions which require a multideterminantal treatment (for example multiplets of atoms, ions,
transition metal complexes, . . . )
Situations in which bonds are broken or partially broken.

Generation of orbitals which are a compromise between the requirements for several states.
Generation of start orbitals for multireference methods covering dynamic correlation (NEVPT2, MRCI,
MREOM,...)
Generation of genuine spin eigenfunctions for multideterminantal/multireference wavefunctions.
In all of these cases the single-determinantal Hartree-Fock method fails badly and in most of these cases DFT
methods will also fail. In these cases a CASSCF method is a good starting point. CASSCF is a special case of
multiconfigurational SCF (MCSCF) methods which specialize to the situation where the orbitals are divided
into three-subspaces: (a) the internal orbitals which are doubly occupied in all configuration state functions
(CSFs), (b) partially occupied (active) orbitals and (c) virtual (external) orbitals which are empty in all CSFs.
A fixed number of electrons is assigned to the internal subspace and the active subspace. If N-electrons are
active in M orbitals one speaks of a CASSCF(N,M) wavefunctions. All spin-eigenfunctions for N-electrons in
M orbitals are included in the configuration interaction step and the energy is made stationary with respect
to variations in the MO and the CI coefficients. Any number of roots of any number of different multiplicities
can be calculated and the CASSCF energy may be optimized with respect to a user defined average of these
states.
The CASSCF method has the nice advantage that it is fully variational which renders the calculation of
analytical gradients relatively easy. Thus, the CASSCF method may be used for geometry optimizations and
numerical frequency calculations.
The price to pay for this strongly enhanced flexibility relative to the single-determinantal HF method is that
the CASSCF method requires more computational ressources and also more insight and planning from the
user side. The technical details are explained in section 9.11. Here we explain the use of the CASSCF method
by examples. In addition to the description in the manual, there is a separate tutorial for CASSCF with
many more examples in the field of coordination chemistry. The tutorial covers the design of the calculation,
practical tips on convergence as well as the computation of properties.
A number of properties are available in ORCA (g-tensor, ZFS splitting, CD, MCD, susceptibility, dipoles,
...). The majority of CASSCF properties such as EPR parameters are computed in the framework of the
quasi-degenerate perturbation theory. Some properties such as ZFS splittings can also be computed via
perturbation theory or rigorously extracted from an effective Hamiltonian. For a detailed description of the
available properties and options see section 9.11.2. All the aforementioned properties are computed within
the CASSCF module. An exception are M ossbauer parameters, which are computed with the usual keywords
using the EPRNMR module (8.13.10).
8.1.7.2 A simple Example
One standard example of a multireference system is the Be atom. Let us run two calculations, a standard
closed-shell calculation (1s2 2s2 ) and a CASSCF(2,4) calculation which also includes the (1s2 2s1 2p1 ) and
(1s2 2s0 2p2 ) configurations.
! TZVPP TightSCF
* xyz 0 1
Be 0 0 0
*
This standard closed-shell calculation yields the energy -14.56213241 Eh. The CASSCF calculation
! TZVPP TightSCF
%casscf nel 2
norb 4
end
* xyz 0 1
Be 0 0 0
*
yields the energy -14.605381525 Eh. Thus, the inclusion of the 2p shell results in an energy lowering of 43
mEh which is considerable. The CASSCF program also prints the composition of the wavefunction:
---------------------------------------------
CAS-SCF STATES FOR BLOCK 1 MULT= 1 NROOTS= 1
---------------------------------------------
ROOT 0: E= -14.6053815294 Eh
0.90060 [ 0]: 2000
0.03313 [ 4]: 0200
0.03313 [ 9]: 0002
0.03313 [ 7]: 0020
This information is to be read as follows: The lowest state is composed of 90% of the configuration which has
the active space occupation pattern 2000 which means that the first active orbital is doubly occupied in this
configuration while the other three are empty. The MO vector composition tells us what these orbitals are
(ORCA uses natural orbitals to canonicalize the active space).
0 1 2 3 4 5
-4.70502 -0.27270 0.11579 0.11579 0.11579 0.16796
2.00000 1.80121 0.06626 0.06626 0.06626 0.00000
-------- -------- -------- -------- -------- --------
0 Be s 100.0 100.0 0.0 0.0 0.0 100.0
0 Be pz 0.0 0.0 13.6 6.1 80.4 0.0
0 Be px 0.0 0.0 1.5 93.8 4.6 0.0
0 Be py 0.0 0.0 84.9 0.1 15.0 0.0
Thus, the first active space orbital has occupion number 1.80121 and is the Be-2s orbital. The other three
orbitals are 2p in character and all have the same occupation number 0.06626. Since they are degenerate
in occupation number space, they are arbitrary mixtures of the three 2p orbitals. It is then clear that the
other components of the wavefunction (each with 3.31%) are those in which one of the 2p orbitals is doubly
occupied.
How did we know how to put the 2s and 2p orbitals in the active space? The answer is WE DID NOT
KNOW! In this case it was good luck that the initial guess produced the orbitals in such an order that we
had the 2s and 2p orbitals active. IN GENERAL IT IS YOUR RESPONSIBILITY THAT THE
ORBITALS ARE ORDERED SUCH THAT THE ORBITALS THAT YOU WANT IN THE
ACTIVE SPACE COME IN THE DESIRED ORDER. In many cases this will require re-ordering
and CAREFUL INSPECTION of the starting orbitals.
ATTENTION:
If you include orbitals in the active space that are nearly empty or nearly doubly occupied, convegence
problems are likely. The SuperCI(PT) and Newton-Raphson method are less prone to these problems.
8.1.7.3 Starting Orbitals
TIP
In many cases natural orbitals of a simple correlated calculation of some kind provide a good starting
point for CASSCF.
Let us illustrate this principle with a calculation on the Benzene molecule where we want to include all six
-orbitals in the active space. After doing a RHF calculation:
! RHF SV(P)
* int 0 1
C 0 0 0 0.000000 0.000 0.000
C 1 0 0 1.389437 0.000 0.000
C 2 1 0 1.389437 120.000 0.000
C 3 2 1 1.389437 120.000 0.000
C 4 3 2 1.389437 120.000 0.000
C 5 4 3 1.389437 120.000 0.000
H 1 2 3 1.082921 120.000 180.000
H 2 1 3 1.082921 120.000 180.000
H 3 2 1 1.082921 120.000 180.000
H 4 3 2 1.082921 120.000 180.000
H 5 4 3 1.082921 120.000 180.000
H 6 5 4 1.082921 120.000 180.000
*
%Output
Print[P_ReducedOrbPopMO_L] 1
End
We can look at the orbitals around the HOMO/LUMO gap:
12 13 14 15 16 17
-0.63810 -0.62613 -0.59153 -0.59153 -0.50570 -0.49833
2.00000 2.00000 2.00000 2.00000 2.00000 2.00000
-------- -------- -------- -------- -------- --------
0 C s 2.9 0.0 0.3 0.1 0.0 0.0
0 C pz 0.0 0.0 0.0 0.0 16.5 0.0
0 C px 1.4 12.4 5.9 0.3 0.0 11.2
0 C py 4.2 4.1 10.1 5.9 0.0 0.1
0 C dyz 0.0 0.0 0.0 0.0 0.1 0.0
0 C dx2y2 0.1 0.1 0.2 0.2 0.0 0.5
0 C dxy 0.4 0.0 0.0 0.2 0.0 0.0
1 C s 2.9 0.0 0.3 0.1 0.0 0.0
1 C pz 0.0 0.0 0.0 0.0 16.5 0.0
1 C px 1.4 12.4 5.9 0.3 0.0 11.2
1 C py 4.2 4.1 10.1 5.9 0.0 0.1
1 C dyz 0.0 0.0 0.0 0.0 0.1 0.0
1 C dx2y2 0.1 0.1 0.2 0.2 0.0 0.5
1 C dxy 0.4 0.0 0.0 0.2 0.0 0.0
2 C s 2.9 0.0 0.0 0.4 0.0 0.1
2 C pz 0.0 0.0 0.0 0.0 16.5 0.0
2 C px 5.7 0.0 0.0 20.9 0.0 10.1
2 C py 0.0 16.5 1.3 0.0 0.0 0.0
2 C dxz 0.0 0.0 0.0 0.0 0.1 0.0
2 C dx2y2 0.6 0.0 0.0 0.2 0.0 1.2
2 C dxy 0.0 0.1 0.5 0.0 0.0 0.0
3 C s 2.9 0.0 0.3 0.1 0.0 0.0
3 C pz 0.0 0.0 0.0 0.0 16.5 0.0
3 C px 1.4 12.4 5.9 0.3 0.0 11.2
3 C py 4.2 4.1 10.1 5.9 0.0 0.1
3 C dyz 0.0 0.0 0.0 0.0 0.1 0.0
3 C dx2y2 0.1 0.1 0.2 0.2 0.0 0.5
3 C dxy 0.4 0.0 0.0 0.2 0.0 0.0
4 C s 2.9 0.0 0.3 0.1 0.0 0.0
4 C pz 0.0 0.0 0.0 0.0 16.5 0.0
4 C px 1.4 12.4 5.9 0.3 0.0 11.2
4 C py 4.2 4.1 10.1 5.9 0.0 0.1
4 C dyz 0.0 0.0 0.0 0.0 0.1 0.0
4 C dx2y2 0.1 0.1 0.2 0.2 0.0 0.5
4 C dxy 0.4 0.0 0.0 0.2 0.0 0.0
5 C s 2.9 0.0 0.0 0.4 0.0 0.1
5 C pz 0.0 0.0 0.0 0.0 16.5 0.0
5 C px 5.7 0.0 0.0 20.9 0.0 10.1
5 C py 0.0 16.5 1.3 0.0 0.0 0.0
5 C dxz 0.0 0.0 0.0 0.0 0.1 0.0
5 C dx2y2 0.6 0.0 0.0 0.2 0.0 1.2
5 C dxy 0.0 0.1 0.5 0.0 0.0 0.0
6 H s 7.5 0.0 7.5 2.5 0.0 2.5

7 H s 7.5 0.0 7.5 2.5 0.0 2.5
8 H s 7.5 0.0 0.0 10.0 0.0 9.9
9 H s 7.5 0.0 7.5 2.5 0.0 2.5
10 H s 7.5 0.0 7.5 2.5 0.0 2.5
11 H s 7.5 0.0 0.0 10.0 0.0 9.9
18 19 20 21 22 23
-0.49833 -0.33937 -0.33937 0.13472 0.13472 0.18198
2.00000 2.00000 2.00000 0.00000 0.00000 0.00000
-------- -------- -------- -------- -------- --------
0 C s 0.1 0.0 0.0 0.0 0.0 2.2
0 C pz 0.0 8.1 24.4 7.8 23.4 0.0
0 C px 0.1 0.0 0.0 0.0 0.0 0.6
0 C py 10.4 0.0 0.0 0.0 0.0 1.7
0 C dxz 0.0 0.4 0.2 0.7 0.7 0.0
0 C dyz 0.0 0.2 0.0 0.7 0.0 0.0
0 C dx2y2 0.0 0.0 0.0 0.0 0.0 0.2
0 C dxy 1.0 0.0 0.0 0.0 0.0 0.5
1 C s 0.1 0.0 0.0 0.0 0.0 2.2
1 C pz 0.0 8.1 24.4 7.8 23.4 0.0
1 C px 0.1 0.0 0.0 0.0 0.0 0.6
1 C py 10.4 0.0 0.0 0.0 0.0 1.7
1 C dxz 0.0 0.4 0.2 0.7 0.7 0.0
1 C dyz 0.0 0.2 0.0 0.7 0.0 0.0
1 C dx2y2 0.0 0.0 0.0 0.0 0.0 0.2
1 C dxy 1.0 0.0 0.0 0.0 0.0 0.5
2 C s 0.0 0.0 0.0 0.0 0.0 2.2
2 C pz 0.0 32.5 0.0 31.2 0.0 0.0
2 C px 0.0 0.0 0.0 0.0 0.0 2.2
2 C py 11.6 0.0 0.0 0.0 0.0 0.0
2 C dxz 0.0 0.1 0.0 0.3 0.0 0.0
2 C dyz 0.0 0.0 0.8 0.0 1.8 0.0
2 C dx2y2 0.0 0.0 0.0 0.0 0.0 0.7
2 C dxy 0.4 0.0 0.0 0.0 0.0 0.0
3 C s 0.1 0.0 0.0 0.0 0.0 2.2
3 C pz 0.0 8.1 24.4 7.8 23.4 0.0
3 C px 0.1 0.0 0.0 0.0 0.0 0.6
3 C py 10.4 0.0 0.0 0.0 0.0 1.7
3 C dxz 0.0 0.4 0.2 0.7 0.7 0.0
3 C dyz 0.0 0.2 0.0 0.7 0.0 0.0
3 C dx2y2 0.0 0.0 0.0 0.0 0.0 0.2
3 C dxy 1.0 0.0 0.0 0.0 0.0 0.5
4 C s 0.1 0.0 0.0 0.0 0.0 2.2
4 C pz 0.0 8.1 24.4 7.8 23.4 0.0
4 C px 0.1 0.0 0.0 0.0 0.0 0.6
4 C py 10.4 0.0 0.0 0.0 0.0 1.7
4 C dxz 0.0 0.4 0.2 0.7 0.7 0.0
4 C dyz 0.0 0.2 0.0 0.7 0.0 0.0
4 C dx2y2 0.0 0.0 0.0 0.0 0.0 0.2
4 C dxy 1.0 0.0 0.0 0.0 0.0 0.5
5 C s 0.0 0.0 0.0 0.0 0.0 2.2
5 C pz 0.0 32.5 0.0 31.2 0.0 0.0
5 C px 0.0 0.0 0.0 0.0 0.0 2.2
5 C py 11.6 0.0 0.0 0.0 0.0 0.0
5 C dxz 0.0 0.1 0.0 0.3 0.0 0.0
5 C dyz 0.0 0.0 0.8 0.0 1.8 0.0

5 C dx2y2 0.0 0.0 0.0 0.0 0.0 0.7
5 C dxy 0.4 0.0 0.0 0.0 0.0 0.0
6 H s 7.4 0.0 0.0 0.0 0.0 11.5
7 H s 7.4 0.0 0.0 0.0 0.0 11.5
8 H s 0.0 0.0 0.0 0.0 0.0 11.5
9 H s 7.4 0.0 0.0 0.0 0.0 11.5
10 H s 7.4 0.0 0.0 0.0 0.0 11.5
11 H s 0.0 0.0 0.0 0.0 0.0 11.5
We see that the occupied -orbitals number 16, 19, 20 and the unoccupied ones start with 21 and 22. However,
the sixth high-lying -orbital cannot easily be found. Thus, let us run a simple selected CEPA/2 calculation
and look at the natural orbitals.
! RHF SV(P)
! moread
%moinp "Test-CASSCF-Benzene-1.gbw"
%mrci citype cepa2

tsel 1e-5
natorbiters 1
newblock 1 *
nroots 1
refs cas(0,0) end
end
end
# ...etc, input of coordinates
The calculation prints the occupation numbers:
N[ 6] = 1.98784765
N[ 7] = 1.98513069
N[ 8] = 1.98508633
N[ 9] = 1.97963799
N[ 10] = 1.97957039
N[ 11] = 1.97737886
N[ 12] = 1.97509724
N[ 13] = 1.97370616
N[ 14] = 1.97360821
N[ 15] = 1.96960145
N[ 16] = 1.96958645
N[ 17] = 1.96958581
N[ 18] = 1.95478929
N[ 19] = 1.91751184
N[ 20] = 1.91747498
N[ 21] = 0.07186879
N[ 22] = 0.07181758
N[ 23] = 0.03203528
N[ 24] = 0.01766832
N[ 25] = 0.01757735
N[ 26] = 0.01708578
N[ 27] = 0.01707675
N[ 28] = 0.01671912
N[ 29] = 0.01526139
N[ 30] = 0.01424982
From these occupation number it becomes evident that there are several natural orbitals which are not quite
doubly occupied MOs. Those with an occupation number of 1.95 and less should certainly be taken as active.
In addition the rather strongly occupied virtual MOs 21-23 should also be active leading to CASSCF(6,6).
Let us see what these orbitals are before starting CASSCF:
! RHF SV(P)
! moread noiter
%moinp "Test-CASSCF-Benzene-2.mrci.nat"
Leading to:
18 19 20 21 22 23
1.00000 1.00000 1.00000 1.00000 1.00000 1.00000
1.95479 1.91751 1.91747 0.07187 0.07182 0.03204
-------- -------- -------- -------- -------- --------
0 C pz 16.5 8.1 24.4 23.4 7.8 16.1
0 C dxz 0.0 0.4 0.2 0.6 0.9 0.1
0 C dyz 0.1 0.2 0.0 0.0 0.6 0.4
1 C pz 16.5 8.1 24.4 23.5 7.8 16.1
1 C dxz 0.0 0.4 0.2 0.6 0.9 0.1
1 C dyz 0.1 0.2 0.0 0.0 0.6 0.4
2 C pz 16.5 32.5 0.0 0.0 31.3 16.3
2 C dxz 0.1 0.1 0.0 0.0 0.2 0.5
2 C dyz 0.0 0.0 0.8 1.9 0.0 0.0
3 C pz 16.5 8.1 24.4 23.4 7.8 16.1
3 C dxz 0.0 0.4 0.2 0.6 0.9 0.1
3 C dyz 0.1 0.2 0.0 0.0 0.6 0.4
4 C pz 16.5 8.1 24.4 23.5 7.8 16.1
4 C dxz 0.0 0.4 0.2 0.6 0.9 0.1
4 C dyz 0.1 0.2 0.0 0.0 0.6 0.4
5 C pz 16.5 32.5 0.0 0.0 31.3 16.3
5 C dxz 0.1 0.1 0.0 0.0 0.2 0.5
5 C dyz 0.0 0.0 0.8 1.9 0.0 0.0
This shows us that these six orbitals are precisely the / orbitals that we wanted to have active (you can
also plot them to get even more insight).
Now we know that the desired orbitals are in the correct order, we can do CASSCF:
! RHF SV(P)
! moread
%casscf nel 6
norb 6
nroots 1
mult 1
switchstep nr
end
Here we have also used a special feature of the CASSCF program the second order convergence provided by
the Newton-Raphson method after a certain convergence has been reached (the switchstep nr statement).
The output of the CASSCF program is:
------------------
CAS-SCF ITERATIONS
------------------
MACRO-ITERATION 1:
--- Inactive Energy E0 = -224.09726054 Eh
CI-ITERATION 0:
-230.5848801034 0.000000000000 ( 0.001) CI-PROBLEM SOLVED
DENSITIES MADE
<<<<<<<<<<<<<<<<<<INITIAL CI STATE CHECK>>>>>>>>>>>>>>>>>>
BLOCK 1 MULT= 1 NROOTS= 1

ROOT 0: E= -230.5848801035 Eh
0.90883 [ 0]: 222000
0.02462 [ 14]: 211110
0.01674 [ 729]: 22020 0.01674971 [ 264]: 2202200
0.01041 [ 5265]: 11011 0.01041170 [ 6350]: 1121101
<<<<<<<<<<<<<<<<<<INITIAL CI STATE CHECK>>>>>>>>>>>>>>>>>>
E(CAS)= -230.5848801034 E DE= 0.000000000

--- Canonicalize Internal Space
--- Canonicalize External Space
--- Energy gap subspaces: Ext-Act = -0.281 Act-Int = -0.001
--- current l-shift: Up(Ext-Act) = 1.88 Dn(Act-Int) = 1.60
N(occ)= 1.96763 1.92238 1.92238 0.07909 0.07909 0.02942
||g|| = 0.109239796 Max(G)= -0.049527078 Rot=45,20
--- Orbital Update [SuperCI(PT)]
--- SX_PT (Skipped TA=0 IT=0): ||X|| = 0.151253839 Max(X)(48,23) = -0.070381658
--- SFit(Active Orbitals)
MACRO-ITERATION 2:
CI-ITERATION 0:
DENSITIES MADE
E(CAS)= -230.590027459 Eh DE= -0.005147355
N(occ)= 1.96232 1.90547 1.9054763.09593498 0593490 0.03
||g|| = 0.032960383 Max(G)= -0.008631611 Rot=40,18
MACRO-ITERATION 3:
CI-ITERATION 0:
DENSITIES MADE
E(CAS)= -230.590266456 Eh DE= -0.000238998
N(occ)= 1.96144 1.90281 1.9028146509852670 0852666 0.03
||g|| = 0.006907681 Max(G)= 0.003101930 Rot=404,18 --- Orbital Update [ SuperCI(PT)]
MACRO-ITERATION 4:
===>>> Convergence to 3.0e-02 achieved - switching to Step=NR
CI-ITERATION 0:
DENSITIES MADE
E(CAS)= -230.590270666 Eh DE= -0.000004209
--- Energy gap subspaces: Ext-Act = 0.244 Act-Int = 0.056
N(occ)= 1.96136 1.90268 1.9026837.09865761 0865759 0.03
||g|| = 0.003412624 Max(G)= 0.000966040 Rot=31,13
--- Orbital Update [ NR]
AUGHESS-ITER 0: E= -0.000000953 <r|r>= 0.000004426
AUGHESS-ITER 1: E= -0.000001593 <r|r>= 0.000000616
AUGHESS-ITER 2: E= -0.000001707 <r|r>= 0.000000095
AUGHESS-ITER 3: E= -0.000001721 <r|r>= 0.000000018
AUGHESS-ITER 4: E= -0.000001725 <r|r>= 0.000000003 => CONVERGED
DE(predicted)= -0.000000862 First Element= 0.999999751
<X(rot)|X(rot)>= 0.000000498
MACRO-ITERATION 5:
CI-ITERATION 0:
DENSITIES MADE
E(CAS)= -230.590271531 Eh DE= -0.000000865

N(occ)= 1.96135 1.90266 1.9026670.098671 0098670 0.0599
||g|| = 0.000057975 Max(G)= 0.0000151393329989,16
---- THE CAS-SCF GRADIENT HAS CONVERGED ----
--- FINALIZING ORBITALS ---
---- DOING ONE FINAL ITERATION FOR PRINTING ----
--- Forming Natural Orbitals
MACRO-ITERATION 6:
--- All densities will be recomputed
CI-ITERATION 0:
DENSITIES MADE
E(CAS)= -230.590271527 Eh DE= 0.000000004
N(occ)= 1.96135 1.90266 1.9026670.098671 0098670 0.0599
||g|| = 0.000057984 Max(G)= 0.000014776 Rot=87,44--------------
CASSCF RESULTS
--------------
Final CASSCF energy : -230.590271527 Eh -6274.6803 eV
First of all you can see how the program cycles between CI-vector optimization and orbital optimization steps
(so-called unfolded two-step procedure). After 4 iterations, the program switches to the Newton-Raphson
solver which then converges very rapidly. In the CASSCF program this produces lengthy integrals and
Hessian files and therefore the size of molecule that can be treated is somewhat limited.
8.1.7.4 CASSCF and Symmetry
The CASSCF program can make some use of symmetry. Thus, it is possible to do the CI calculations
separated by irreducible representations. This allows one to calculate electronic states in a more controlled
fashion.
Let us look at a simple example: C2 H4 . We first generate symmetry adapated MP2 natural orbitals. Since
we opt for initial guess orbitals, the computationally cheaper unrelaxed density suffices:
! def2-TZVP def2-TZVP/C UseSym RI-MP2 conv # conventional is faster for small molecules
%mp2
density unrelaxed
natorbs true
end
* int 0 1
C 0 0 0 0 0 0
C 1 0 0 1.35 0 0
H 1 2 0 1.1 120 0
H 1 2 3 1.1 120 180
H 2 1 3 1.1 120 0
H 2 1 3 1.1 120 180
*
The program does the following. It first identifies the group correctly as D2h and sets up its irreducible repre-
sentations. The process detects symmetry within SymThresh (104 ) and purifies the geometry thereafter:
------------------
SYMMETRY DETECTION
------------------
Preparing Data ... done
Detection Threshold: SymThresh ... 1.0000e-04
Point Group will now be determined:

Moving molecule to center of mass ... done
POINT GROUP ... D2h
The coordinates will now be cleaned:

Moving to standard coord frame ... done
(Changed the main axis to z and one of the C2s to x)
Structure cleanup requested ... yes
Selected point group ... D2h
Cleaning Tolerance SymThresh ... 1.0000e-04
Some missing point group data is constructed:

Constructing symmetry operations ... done
Creating atom transfer table ... done
Creating asymmetric unit ... done
Cleaning coordinates ... done
-----------------------------------------------
SYMMETRY-PERFECTED CARTESIAN COORDINATES (A.U.)
-----------------------------------------------
0 C 1.27556514 0.00000000 0.00000000
1 C -1.27556514 0.00000000 0.00000000
2 H 2.31491451 1.80020592 0.00000000
3 H 2.31491451 -1.80020592 0.00000000
4 H -2.31491451 1.80020592 0.00000000
5 H -2.31491451 -1.80020592 0.00000000
------------------
SYMMETRY REDUCTION
------------------
ORCA supports only abelian point groups.
It is now checked, if the determined point group is supported:
Point Group ( D2h ) is ... supported
(Re)building abelian point group:

Creating Character Table ... done
Making direct product table ... done
----------------------
ASYMMETRIC UNIT IN D2h
----------------------
# AT MASS COORDS (A.U.) BAS
0 C 12.0110 1.27556514 0.00000000 0.00000000 0
2 H 1.0080 2.31491451 1.80020592 0.00000000 0
----------------------
SYMMETRY ADOPTED BASIS
----------------------
The coefficients for the symmetry adopted linear combinations (SALCS)
of basis functions will now be computed:
Number of basis functions ... 86
Preparing memory ... done
Constructing Gamma(red) ... done
Reducing Gamma(red) ... done
Constructing SALCs ... done
Checking SALC integrity ... nothing suspicious
Normalizing SALCs ... done
Storing the symmetry object:

Symmetry file ... Test-SYM-CAS-C2H4-1.sym.tmp
Writing symmetry information ... done
It then performs the SCF calculation and keeps the symmetry in the molecular orbitals.

0 2.0000 -11.236728 -305.7669 1-Ag
1 2.0000 -11.235157 -305.7242 1-B3u
2 2.0000 -1.027144 -27.9500 2-Ag
3 2.0000 -0.784021 -21.3343 2-B3u
4 2.0000 -0.641566 -17.4579 1-B2u
5 2.0000 -0.575842 -15.6694 3-Ag
6 2.0000 -0.508313 -13.8319 1-B1g
7 2.0000 -0.373406 -10.1609 1-B1u
8 0.0000 0.139580 3.7982 1-B2g
9 0.0000 0.171982 4.6799 4-Ag
10 0.0000 0.195186 5.3113 3-B3u
11 0.0000 0.196786 5.3548 2-B2u
12 0.0000 0.242832 6.6078 2-B1g
13 0.0000 0.300191 8.1686 5-Ag
14 0.0000 0.326339 8.8801 4-B3u
... etc
The MP2 module does not take any advantage of this information but produces natural orbitals that are
symmetry adapted:
N[ 0](B3u) = 2.00000360
N[ 1]( Ag) = 2.00000219
N[ 2]( Ag) = 1.98056435
N[ 3](B3u) = 1.97195041
N[ 4](B2u) = 1.96746753
N[ 5](B1g) = 1.96578954
N[ 6]( Ag) = 1.95864726
N[ 7](B1u) = 1.93107098
N[ 8](B2g) = 0.04702701
N[ 9](B3u) = 0.02071784
N[ 10](B2u) = 0.01727252
N[ 11]( Ag) = 0.01651489
N[ 12](B1g) = 0.01602695
N[ 13](B3u) = 0.01443373
N[ 14](B1u) = 0.01164204
N[ 15]( Ag) = 0.01008617
N[ 16](B2u) = 0.00999302
N[ 17]( Ag) = 0.00840326
N[ 18](B3g) = 0.00795053
N[ 19](B3u) = 0.00532044
N[ 20]( Au) = 0.00450556
etc.
From this information and visual inspection you will know what orbitals you will have in the active space:
These natural orbitals can then be fed into the CASSCF calculation. We perform a simple calculation in
which we keep the ground state singlet (A1g symmetry, irrep=0) and the first excited triplet state (B3u
symmetry, irrep=7). In general the ordering of irreps follows standard conventions and in case of doubt you
will find the relevant number for each irrep in the output.
For example, here (using LargePrint):
----------------------------
CHARACTER TABLE OF GROUP D2h
----------------------------
GAMMA O1 O2 O3 O4 O5 O6 O7 O8
Ag : 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
B1g: 1.0 1.0 -1.0 -1.0 1.0 1.0 -1.0 -1.0
B2g: 1.0 -1.0 1.0 -1.0 1.0 -1.0 1.0 -1.0
B3g: 1.0 -1.0 -1.0 1.0 1.0 -1.0 -1.0 1.0
Au : 1.0 1.0 1.0 1.0 -1.0 -1.0 -1.0 -1.0
B1u: 1.0 1.0 -1.0 -1.0 -1.0 -1.0 1.0 1.0
B2u: 1.0 -1.0 1.0 -1.0 -1.0 1.0 -1.0 1.0
B3u: 1.0 -1.0 -1.0 1.0 -1.0 1.0 1.0 -1.0
---------------------------------
DIRECT PRODUCT TABLE OF GROUP D2h
---------------------------------
** Ag B1g B2g B3g Au B1u B2u B3u
Ag Ag B1g B2g B3g Au B1u B2u B3u

B1g B1g Ag B3g B2g B1u Au B3u B2u
B2g B2g B3g Ag B1g B2u B3u Au B1u
B3g B3g B2g B1g Ag B3u B2u B1u Au
Au Au B1u B2u B3u Ag B1g B2g B3g

B1u B1u Au B3u B2u B1g Ag B3g B2g
B2u B2u B3u Au B1u B2g B3g Ag B1g
B3u B3u B2u B1u Au B3g B2g B1g Ag
We use the following input for CASSCF, where we tightened the integral cut-offs and the the convergence
criteria using !VeryTightSCF.
! def2-TZVP Conv NormalPrint UseSym

! moread
%moinp "Test-SYM-CAS-C2H4-1.mp2nat"
%casscf nel 4
norb 4
# This is only here to show that NR can also be used from
# the start with orbstep
orbstep nr
switchstep nr
# the lowest singet and triplet states. The new feature
# is the array "irrep" that lets you give the irrep for
# a given block. Thus, now you can have several blocks of
# the same multiplicity but different spatial symmetry
irrep 0,7
mult 1,3
nroots 1,1
end
* int 0 1
C 0 0 0 0 0 0
C 1 0 0 1.35 0 0
H 1 2 0 1.1 120 0
H 1 2 3 1.1 120 180
H 2 1 3 1.1 120 0
H 2 1 3 1.1 120 180
*
And gives:
------------
SCF SETTINGS
------------
Hamiltonian:
Ab initio Hamiltonian Method .... Hartree-Fock(GTOs)
General Settings:
Integral files IntName .... Test-SYM-CAS-C2H4-1
Hartree-Fock type HFTyp .... CASSCF
Total Charge Charge .... 0

Multiplicity Mult .... 1
Number of Electrons NEL .... 16
Basis Dimension Dim .... 86
Nuclear Repulsion ENuc .... 32.9609050695 Eh
Symmetry handling UseSym .... ON

Point group .... D2h
Used point group .... D2h
Number of irreps .... 8
Irrep Ag has 19 symmetry adapted basis functions (ofs= 0)
Irrep B1g has 12 symmetry adapted basis functions (ofs= 19)
Irrep Au has 4 symmetry adapted basis functions (ofs= 43)
Irrep B1u has 8 symmetry adapted basis functions (ofs= 47)
And further in the CASCSF program:
Symmetry handling UseSym ... ON

Point group ... D2h
Used point group ... D2h
Number of irreps ... 8
Irrep Ag has 19 SALCs (ofs= 0) #(closed)= 2 #(active)= 1
Irrep B1g has 12 SALCs (ofs= 19) #(closed)= 1 #(active)= 0
Irrep Au has 4 SALCs (ofs= 43) #(closed)= 0 #(active)= 0
Irrep B1u has 8 SALCs (ofs= 47) #(closed)= 0 #(active)= 1
Symmetries of active orbitals:
MO = 6 IRREP= 0 (Ag)
MO = 7 IRREP= 5 (B1u)
MO = 8 IRREP= 2 (B2g)
MO = 9 IRREP= 7 (B3u)
Setting up the integral package ... done

Building the CAS space ... done (7 configurations for Mult=1 Irrep=0)
Building the CAS space ... done (4 configurations for Mult=3 Irrep=7)
Note that the irrep occupations and active space irreps will be frozen to what they are upon entering the
CASSCF program. This helps to avoid flipping electronic states or active orbitals in and out of the active
space.
After which it smoothly converges to give:
6: 1.986258 -0.753012 -20.4905 3-Ag

7: 1.457849 -0.291201 -7.9240 1-B1u
8: 0.541977 0.100890 2.7454 1-B2g
9: 0.013915 0.964186 26.2368 3-B3u
As well as:
-----------------------------
SA-CASSCF TRANSITION ENERGIES
------------------------------
LOWEST ROOT = -78.110314788 Eh -2125.490 eV
STATE ROOT MULT IRREP DE/a.u. DE/eV DE/cm**-1

1: 0 3 B3u 0.163741 4.456 35937.1
8.1.7.5 RI, RIJCOSX and RIJK approximations for CASSCF
A significant speedup of CASSCF calculations on larger molecules can be achieved with the RI and RIJCOSX
approximation. For CASSCF calculations, we the keyword trafostep RI and specify an additional auxiliary
basis (/c or the more accurate /jk auxiliary basis).
! RHF SV(P) def2-svp/C

! moread
%casscf nel 6
norb 6
nroots 1
mult 1
switchstep nr
trafostep ri
end
The energy of this calculation is -230.590328 Eh compared to the previous result -230.590271 Eh. Thus,
the RI error is only 0.06 mEh which is certainly negligible for all intents and purposes. With the larger /jk
auxiliary basis the error is typically much smaller (0.02 mEh in this example). Even if more accurate results
are necessary, it is a good idea to pre-converge the CASSCF with RI. The resulting orbitals should be a much
better guess for the subsequent calculation without RI and thus save computation time.
The TrafoStep ri only affects the integral transformation in CASSCF calculations while the Fock operators
are still calculated in the standard way using four index integrals. In order to fully avoid any four-index
integral evaluation you can significantly speed up the time needed in each iteration by specifying ! RIJCOSX
. The keyword implies TrafoStep rimo. The COSX approximation is used for the construction of the Fock
matrices. In this case, the /jk auxiliary basis is mandatory.
! RHF SV(P) RIJCOSX def2/jk

! moread
%casscf nel 6
norb 6
nroots 1
mult 1
end
The speedup and accuracy is similar to what is observed in RHF and UHF calculations. In this example
the RIJCOSX leads to an error of 1 mEh. The methodology performs better for the computation of energy
differences, where it profits from error cancellation. The RIJCOSX is ideally suited to converge large-scale
systems. Note that for large calculations the integral cut-offs and numerical grids should be tightened. See
section 9.2.2.6 for details. With a floppy numerical grid setting the accuracy as well as the convergence
behavior of CASSCF deteriorate. For systems that are large but still feasible for a conventional calculation it
is recommended to use the RIJK approximations instead (!RIJK conv).
! RHF SV(P) RIJK conv def2/jk
The RIJK methodology is more accurate and robust for CASSCF e.g. here the error is just 0.5 mEH. Organic
molecules with nearly double occupied orbitals can be challenge for the orbital optimization process. We
compare to a calculation done without the NR solver:
! RHF SV(P)
! moread
%casscf nel 6
norb 6
nroots 1
mult 1
end
This calculation takes 8 cycles to converge. The default orbital optimization uses a first order method
(SuperCI PT), which might take more iterations compared to the NR method. However, first order methods
are much cheaper than the NRNand therefore it may pay off to do a few iterations more rather than switching
to the expensive second order methods. On top of that, second order methods are also less robust and may
diverge in certain circumstances. When playing with the convergence settings, there is always a trade-off
between speed versus robustness. The default settings are slightly more aggressive (faster). When facing
convergence problems, it can be useful to use a more robust scheme (orbstep SuperCI and switchstep
DIIS).
8.1.7.6 Breaking Chemical Bonds
Let us turn to the breaking of chemical bonds. The first (trivial) example. Note that TightSCF tightens the
convergence settings of CASSCF. is H2 :
!SVP TightSCF
%casscf nel 2
norb 2
mult 1
nroots 1
# disable abort on active space changes
actconstraints 0
end
%paras R [0.6 0.65 0.70 0.75 0.80 0.90 1.0 1.1 1.3 1.5 1.7
2.0 2.2 2.4 2.6 2.9 3.2 3.5 3.8 4.1]
end
* xyz 0 1
h 0 0 0
h 0 0 {R}
*
There is a small complication in the input above. By default CASSCF runs an active space protection scheme
(actconstraints), which aborts if the active spaces changes its composition. In many situations this is a
good idea. For this specific parameter scan, the feature is counter-productive. In the initial geometry (0.6
A)
the active space orbitals have a large contributions from the pz orbitals (50% for the antibonding orbital).
# Loewdin population analysis of the active orbitals at 0.6

A
0 1
-0.61543 0.73178
1.98225 0.01775
-------- --------
0 H s 48.5 25.0
0 H pz 1.5 25.0
0 H px 0.0 0.0
0 H py 0.0 0.0
1 H s 48.5 25.0
1 H pz 1.5 25.0
1 H px 0.0 0.0
1 H py 0.0 0.0
The converged orbitals are then used in the next geometry. However, as the hydrogens atoms separate the pz
contribution vanishes, which will by default trigger a program abort since the composition of the active space
has changed. More information on this can be found in section 9.11.To avoid an uncessary program abort, we
added actconstraints 0 in the input to disable the feature.
The resulting potential energy surface (PES) is depicted in 8.10 together with PESs obtained from RHF and
broken-symmetry UHF calculations (input below).
! RHF SVP TightSCF

# etc...
And
! UHF SVP TightSCF
%scf FlipSpin 1 FinalMs 0.0

end
Figure 8.10: Potential Energy Surface of the H2 molecule from RHF, UHF and CASSCF(2,2) cal-
culations (SVP basis).
It is obvious, that the CASSCF surface is concise and yields the correct dissociation behavior. The RHF
surface is roughly parallel to the CASSCF surface in the vicinity of the minimum but then starts to fail badly
as the H-H bond starts to break. The broken-symmetry UHF solution is identical to RHF in the vicinity of
the minimum and dissociates correctly. It is, however, of rather mediocre quality in the intermediate region
where it follows the RHF surface too long too closely.
A more challenging case is to dissociate the N-N bond of the N2 molecule correctly. Using CASSCF with the
six p-orbitals we get a nice potential energy curve (The depth of the minimum is still too shallow compared
to experiment by some 1 eV or so. A good dissociation energy requires a dynamic correlation treatment on
top of CASSCF and a larger basis set).
One can use the H2 example to illustrate the state-averaging feature. Since we have two active electrons we
have two singlets and one triplet. Let us average the orbitals over these three states (we take equal weights
for all multiplicity blocks):
Figure 8.11: Potential Energy Surface of the N2 molecule from CASSCF(6,6) calculations (SVP
basis).
!SVP TightSCF
%casscf nel 2
norb 2
mult 1,3
nroots 2,1
# weighting below corresponds the program default
# and does not need to be specified explicitly
bweight = 1,1 #equal weights per multiplicity blocks.
weights[0] = 1,1 #equal weights within a given multiplicity block
weights[1] = 1
actconstraints 0 # disabled
end
which gives:
Figure 8.12: State averaged CASSCF(2,2) calculations on H2 (two singlets, one triplet; SVP basis).
The grey curve is the ground state CASSCF(2,2) curve
One observes, that the singlet and triplet ground states become degenerate for large distances (as required)
while the second singlet becomes the ionic singlet state which is high in energy. If one compares the lowest
root of the state-averaged calculation (in green) with the dedicated ground state calculation (in grey) one
gets an idea of the energetic penalty that is associated with averaged as opposed to dedicated orbitals.
A more involved example is the rotation around the double bond in C2 H4 . Here, the -bond is broken as one
twists the molecule. The means the proper active space consists of two active electron in two orbitals.
The input is (for fun, we average over the lowest two singlets and the triplet):
!SV(P) SV/C SmallPrint NoPop NoMOPrint
%casscf nel = 2
norb = 2
mult = 1,3
nroots = 2,1
bweight = 2,1
weights[0] = 1,1
weights[1] = 1
TrafoStep RI
end
%paras R= 1.3385
Alpha=0,180,37
end
* int 0 1
C 0 0 0 0 0 0
C 1 0 0 {R} 0 0
H 1 2 0 1.07 120 0
H 1 2 3 1.07 120 180
H 2 1 3 1.07 120 {Alpha}
H 2 1 3 1.07 120 {Alpha+180}
*
Figure 8.13: State averaged CASSCF(2,2) calculations on C2 H4 (two singlets, one triplet; SV(P)
basis). The grey curve is the state averaged energy and the purple curve corresponds
to RHF.
We can see from this plot, that the CASSCF method produces a nice ground state surface with the correct
periodicity and degeneracy at the end points, which represent the planar ethylene molecule. At 90 one has
a weakly coupled diradical and the singlet and triplet states become nearly degenerate, again as expected.
Calculations with larger basis sets and inclusion of dynamic correlation would give nice quantitative results.
We have also plotted the RHF energy (in purple) which gives a qualitatively wrong surface and does not
return to the correct solution for planar ethylene. It is evident that even high quality dynamic correlation
treatments like CC or CI would hardly be able to repair the shortcomings of the poor RHF reference state.
In all these cases, CASSCF is the proper starting point for higher accuracy.
8.1.7.7 Excited States
As a final example, we do a state-average calculation on H2 CO in order to illustrate excited state treatments.

We expect from the ground state (basically closed-shell) a n and a excited state which we want
to describe. For the n we also want to calculate the triplet since it is well known experimentally. First
we take DFT orbitals as starting guess.
! BP86 ma-def2-SVP TightSCF

%base "1"
*int 0 1
C 0 0 0 0.00 0.0 0.00
O 1 0 0 1.20 0.0 0.00
H 1 2 0 1.10 120.0 0.00
H 1 2 3 1.10 120.0 180.00
*
$new_job
! ma-def2-SVP TightSCF
! moread
%moinp "1.gbw"
%base "Test-CASSCF.H2CO-1"
%casscf nel 4
norb 3
switchstep nr
mult 1,3
nroots 3,1
end
*int 0 1
C 0 0 0 0.00 0.0 0.00
O 1 0 0 1.20 0.0 0.00
H 1 2 0 1.10 120.0 0.00
H 1 2 3 1.10 120.0 180.00
*
We get:
-----------------------------
------------------------------
LOWEST ROOT = -113.736217779 Eh -3094.842 eV
STATE ROOT MULT DE/a.u. DE/eV DE/cm**-1

1: 0 3 0.056271 1.531 12349.9
2: 1 1 0.075355 2.050 16538.5
3: 2 1 0.272589 7.417 59826.3
Which is very poor. So, here the treatment based on average orbitals does not work at all. So let us do the
states one after the other by adjusting the weights:
Ground state:
%casscf nel 4
norb 3
switchstep nr
mult 1,3
nroots 3,1
bweight = 1,0
weights[0] = 1,0,0
weights[1] = 0
end
First-Singlet:
%casscf nel 4
norb 3
mult 1,3
nroots 3,1
bweight = 1,0
weights[0] = 0,1,0
weights[1] = 0
end
Second-Singlet:
%casscf nel 4
norb 3
mult 1,3
nroots 3,1
bweight = 1,0
weights[0] = 0,0,1
weights[1] = 0
end
First-Triplet
%casscf nel 4
norb 3
mult 1,3
nroots 3,1
bweight = 0,1
weights[0] = 0,0,0
weights[1] = 1
end
The total energies are:
Ground State: -113.825085148

First-Singlet -113.671020811 = 4.19 eV
Second-Singlet -113.479355179 = 9.41 eV
First Triplet -113.684545957 = 3.82 eV
These results are much better and as soon as we include dynamic correlation one would have good transition
energies (future releases will include state specific perturbation corrections to achieve this). This calculation
shows the limits of state averaging if the orbitals strongly change (as in the present case, since n
orbitals are very different from orbitals) the state averaging does not lead to good transition energies.
In this case, one should follow SA-CASSCF by NEVPT2/MRCI/SORCI (see below) or alternatively use
state specific procedures to optimize each state in turn and then calculate its correlation energy. However,
converging CASSCF to an excited state can be a challenge with the two-step orbital optimization procedures
implemented in ORCA .
8.1.7.8 CASSCF Natural Orbitals as Input for Coupled-Cluster Calculations
Consider the possibility that you are not sure about the orbital occupancy of your system. Hence you
carry out some CASSCF calculation for various states of the system in an effort to decide on the ground
state. You can of course follow the CASSCF by MR-MP2 or MR-ACPF or SORCI calculations to get a true
multireference result for the state ordering. Yet, in some cases you may also want to obtain a coupled-cluster
estimate for the state energy difference. Converging coupled-cluster calculation on alternative states in
a controlled manner is anything but trivial. Here a feature of ORCA might be helpful. The best single
configuration that resembles a given CASSCF state is built from the natural orbitals of this state. These
orbitals are also ordered in the right way to be input into the MDCI program. The convergence to excited
states is, of course, not without pitfalls and limitations as will become evident in the two examples below.
As a negative example consider first the following interesting dicarbene.
For this molecule we expect that we should use four active orbitals and four active electrons and that singlet,
triplet and quintet states might be accessible. We start with a simple CASSCF(4,4) optimization on the
lowest singlet state.
! SV(P) Conv TightSCF Opt
%casscf nel 4 norb 4 end
* int 0 1
C 0 0 0 0 0 0
C 1 0 0 1.45 0 0
C 2 1 0 1.45 109.4712 0
H 1 2 3 1.1 109.4712 0
H 3 2 1 1.1 109.4712 0
H 2 1 3 1.1 109.4712 240
H 2 1 3 1.1 109.4712 120
*
Of course, one should also do optimizations on the the other two spin states (and with larger basis sets) but
for the sake of the argument, we stick to the singlet structure. Next, the natural orbitals for each state are
generated with the help of the MRCI module. To this end we run a state averaged CASSCF for the lowest
four singlet, two triplets and the quintet and pass that information on to the MRCI module that does a
CASCI only (e.g. no excitations):
! ano-pVDZ Conv TightSCF MRCI

%casscf nel 4
norb 4
mult 1,3,5
nroots 4,2,1
end
%mrci tsel 0
tpre 0
donatorbs 2
densities 5,1
newblock 1 * nroots 4 excitations none refs cas(4,4) end end
end
* int 0 1
C 0 0 0 0.000000000000 0.00000000 0.00000000
C 1 0 0 1.494834528132 0.00000000 0.00000000
C 2 1 0 1.494834528211 105.15548921 0.00000000
H 1 2 3 1.083843964350 129.42964540 0.00000000
H 3 2 1 1.083843964327 129.42964555 0.00000000
H 2 1 3 1.094075308221 111.18220523 239.57277074
H 2 1 3 1.094075308221 111.18220523 120.42722926
*
This produces the files: BaseName.bm sn.nat where m is the number of the block (m=0,1,2 correspond to
singlet, triplet and quintet respectively) and n stands for the relevant state (n=0,1,2,3 for singlet, n=0,1
for triplet and n=0 for quintet).
These natural orbitals are then fed into unrestricted QCISD(T) calculations:
! UHF ano-pVDZ Conv TightSCF AOX-QCISD(T)

! moread noiter
%moinp "C05S01_101.b0_s0.nat"
* int 0 1
C 0 0 0 0.000000000000 0.00000000 0.00000000
C 1 0 0 1.494834528132 0.00000000 0.00000000
C 2 1 0 1.494834528211 105.15548921 0.00000000
H 1 2 3 1.083843964350 129.42964540 0.00000000
H 3 2 1 1.083843964327 129.42964555 0.00000000
H 2 1 3 1.094075308221 111.18220523 239.57277074
H 2 1 3 1.094075308221 111.18220523 120.42722926
*
This produces the energies:
State Energy (Eh) Relative Energy (cm1 )

S0 -116.190768 0
S1 -116.067138 27133.0
S2 -116.067138 27133.0
S3 -116.067138 27133.0
T0 -116.155421 7757.6
T1 -116.113969 16855.1
Q0 -116.134575 12332.6
It is found that the calculations indeed converge to different states. The excited singlets all fall down to the
same state that is approximately 27,000 cm1 above the lowest solution. The triplets are distinct and the
quintet is unique anyways. Inspection of the coupled-cluster wavefunctions indicate that the singlet converged
to the closed-shell solution and the first doubly excited state respectively.
These energies can be compared with the genuine multireference calculation obtained from the SORCI
method:
! ano-pVDZ Conv TightSCF SORCI
%casscf nel 4
norb 4
mult 1,3,5
nroots 4,2,1
end
* int 0 1
C 0 0 0 0.000000000000 0.00000000 0.00000000
C 1 0 0 1.494834528132 0.00000000 0.00000000
C 2 1 0 1.494834528211 105.15548921 0.00000000
H 1 2 3 1.083843964350 129.42964540 0.00000000
H 3 2 1 1.083843964327 129.42964555 0.00000000
H 2 1 3 1.094075308221 111.18220523 239.57277074
H 2 1 3 1.094075308221 111.18220523 120.42722926
*
Which produces:
State Mult Irrep Root Block mEh eV 1/cm

0 1 -1 0 2 0.000 0.000 0.0
1 3 -1 0 1 1.355 0.037 297.3
2 5 -1 0 0 26.113 0.711 5731.1
3 3 -1 1 1 41.126 1.119 9026.1
4 1 -1 2 2 79.540 2.164 17457.0
5 1 -1 1 2 84.407 2.297 18525.3
6 1 -1 3 2 86.175 2.345 18913.3
With the description of the wavefunctions:

Singlets:
STATE 0: Energy= -115.944916420 Eh RefWeight= 0.9224 0.00 eV 0.0 cm**-1
0.3547 : h---h---[2020]
0.3298 : h---h---[2002]
0.1034 : h---h---[1111]
0.0681 : h---h---[0220]
0.0663 : h---h---[0202]
0.6769 : h---h---[2200]
0.0638 : h---h---[2020]
0.0710 : h---h---[2002]
0.0877 : h---h---[1111]
0.0039 : h---h---[0220]
0.0051 : h---h---[0202]
0.0055 : h---h---[0022]
0.7789 : h---h---[2110]
0.0920 : h---h---[1201]
0.0149 : h---h---[1021]
0.0112 : h---h---[0112]
0.0038 : h---h 6[2120]
0.0049 : h---h---[2100]p14
0.0036 : h---h---[1110]p13
0.7580 : h---h---[2101]
0.1089 : h---h---[1210]
0.0221 : h---h---[1012]
0.0098 : h---h---[0121]
0.0064 : h---h 6[2111]
0.0037 : h---h---[1101]p13
Triplets:
0.6638 : h---h---[2011]
0.0675 : h---h---[1120]
0.0651 : h---h---[1102]
0.1246 : h---h---[0211]
0.6861 : h---h---[2110]
0.1914 : h---h---[1201]
0.0244 : h---h---[1021]
0.0152 : h---h---[0112]
Quintets:
0.9263 : h---h---[1111]
Thus, the singlet ground state is heavily multiconfigurational, the lowest triplet state is moderately multicon-
figurational and the lowest quintet state is of course a single configuration. Interstingly, the lowest singlet,
triplet and quintet do not form a regular spin ladder which might have been expected if one considers the
system of being composed of two interacting S=1 systems. Rather, the lowest singlet and triplet states are
close in energy while the lowest quintet is far away.
The energies are completely different from the QCISD(T) results. However, this is not unexpected based on
the composition of these wavefunctions. These are the limitations of single reference methods. Nevertheless,
this shows how such results can be obtained in principle.
As a more positive example, consider some ionized states of the water cation:
First the natural orbital generation:
! ano-pVDZ Conv TightSCF
%casscf nel 7
norb 6
nroots 3
end
%mrci tsel 0
tpre 0
donatorbs 2
densities 5,1
newblock 2 *
nroots 3
excitations none
refs cas(7,6)end
end
end
* int 1 2
O 0 0 0 0.000000 0.000 0.000
H 1 0 0 1.012277 0.000 0.000
H 1 2 0 1.012177 109.288 0.000
*
Then the SORCI reference calculation:
! ano-pVDZ Conv TightSCF SORCI
%casscf nel 7
norb 6
nroots 3
end
* int 1 2
O 0 0 0 0.000000 0.000 0.000
H 1 0 0 1.012277 0.000 0.000
H 1 2 0 1.012177 109.288 0.000
*
Then the three QCISD(T) calculations

! UHF ano-pVDZ Conv TightSCF QCISD(T)

! moread noiter
%moinp "H2O+-02.b0_s0.nat"
* int 1 2
O 0 0 0 0.000000 0.000 0.000
H 1 0 0 1.012277 0.000 0.000
H 1 2 0 1.012177 109.288 0.000
*
we obtain the transition energies:
SORCI QCISD(T) (in cm-1)

D0 0 0.0
D1 16269 18283
D2 50403 50509
Thus, in this example the agreement between single- and multireference methods is good and the unrestricted
QCISD(T) method is able to describe these excited doublet states. The natural orbitals have been a reliable
way to guide the CC equations into the desired solutions. This will work in many cases.
8.1.8 N-Electron Valence State Perturbation Theory (NEVPT2)
NEVPT2 is an internally contracted multireference perturbation theory, which applies to CASSCF type
wavefunctions. The NEVPT2 method as described in the original papers of Angeli et al comes in two
flavor the strongly contracted NEVPT2 (SC-NEVPT2) and the so called partially contracted NEVPT2
(PC-NEVPT2). [116118] In fact, the latter employs a fully internally contracted wavefunction and should
more appropriately called FIC-NEVPT2. Both methods produces energies of similar quality as the CASPT2
approach. [119, 120]
The SC-NEVPT2 and FIC-NEVPT2 are implemented in ORCA together with a number of approximations
that makes the methodology very attractive for large scale applications. In conjunction with the RI
approximation systems with active space of to 16 active orbitals and 2000 basis functions can be computed.
With the newly developed DLPNO version of the FIC-NEVPT2 the size of the molecules does not matter
anymore. [121] For a more complete list of keywords and features we refer to detailed documation section
9.12.
Besides corrections to the correlation energy, ORCA features UV, IR, CD and MCD spectra as well as
EPR parameters for NEVPT2. These properties are computed using the quasi-degenerate perturbation
theory that is described in section 9.11.2. The NEVPT2 corrections enter as improved diagonal energies
in this formalism. ORCA also features the multi-state extension (QD-NEVPT2) for the strongly contracted
NEVPT2 variant.
As a simple example consider the ground state of the nitrogen molecule N2 . After defining the computational
details of our CASSCF calculation, we insert !NEVPT2 as simple input or specify NEVPT2 SC in the
%casscf block. The SC-NEVPT2 has been the workhorse of our own group for a long time and is thus the
program default.
!def2-svp nofrozencore
%casscf nel 6
norb 6
mult 1
NEVPT2 SC # SC for the strongly contracted NEVTP2
# FIC for the fully internally contracted NEVPT2
# DLPNO for the FIC-NEVPT2 with DLPNO.
# DLPNO requires: trafostep RI and an aux basis
end
* xyz 0 1
N 0.0 0.0 0.0
N 0.0 0.0 1.09768
*
For better control of the program flow it is advised to split the calculation into two parts. First converge
the CASSCF calculation and then in a second step read the converged orbitals and execute the actual
NEVPT2.
---------------------------------------------------------------
ORCA-CASSCF
---------------------------------------------------------------
...
NEVPT2-SETTINGS:
A NEVPT2 calculation will be performed on top of the CASSCF wavefunction!
...
---------------------------------------------------------------
< NEVPT2 >
---------------------------------------------------------------
...
===============================================================
NEVPT2 Results
===============================================================
*********************
MULT 1, ROOT 0
*********************
Class V0_ijab : dE = -0.017748

Class Vm1_iab : dE = -0.023171

Class Vm2_ab : dE = -0.042194
Class V1_ija : dE = -0.006806
Class V2_ij : dE = -0.005056
Class V0_ia : dE = -0.054000
Class Vm1_a : dE = -0.007091
Class V1_i : dE = -0.001963
---------------------------------------------------------------
Total Energy Correction : dE = -0.15802909
---------------------------------------------------------------
Zero Order Energy : E0 = -108.98888640
---------------------------------------------------------------
Total Energy (E0+dE) : E = -109.14691549
---------------------------------------------------------------
Introducing dynamic correlation with NEVPT2 lowers the energy by 150 mEh. ORCA also prints the
contribution of each excitation class V to the NEVPT2 correction. We note that in the case of a single
reference wavefunction the V0 ij,ab excitation class produces the exact MP2 correlation energy. Different
from the ORCA 3.0, the default setting of NEVPT2 calculation uses the frozen core approximation. If one
need to reproduce the NEVPT2 energies of ORCA 3.0 calculation, the keyword nofrozencore should be
added. In our opinion, the frozen core should be avoided. The savings in computation time are fairly small
and do not justify the loss in accuracy.
In chapter 8.1.7.6 the dissociation of the N2 molecule has been investigated with the CASSCF method.
Inserting NEVPT2 SC into the %casscf block we obtain the NEVPT2 correction as additional information.
!def2-svp nofrozencore
%casscf nel 6
norb 6
mult 1
nevpt2 sc
end
%paras
R = 0.7, 2.5, 30
end
*xyz 0 1
N 0.0 0.0 0.0
N 0.0 0.0 {R}
*
All of the options available in CASSCF can in principle be applied to NEVPT2. Since NEVPT2 is implemented
Figure 8.14: Potential Energy Surface of the N2 molecule from CASSCF(6,6) and NEVPT2 calcu-
lations (def2-SVP).
as a submodule of CASSCF, it will inherit all settings from CASSCF (!tightscf, !UseSym, !RIJCOSX,
. . . ).
NOTE
NEVPT2 analytic gradients are not available, but numerical gradients are!
8.1.9 Full Configuration Interaction Energies and Properties
The MRCI and CASSCF modules may be used to produce energies and properties (g-tensors, zero-field
splitting, etc) at the level of full configuration interaction (FCI). Larger molcular system naturally require
more computational resources and time. For the CASSCF module, the limit of still feasible calculations is
around 14 electrons in 14 orbitals using configuration state functions (CSF) for the CI expansion. Larger
active spaces may be computed in the framework of the Density Matrix Renormalization Group (DMRG),
which may be selected by CIStep in the CASSCF module. Similarly, the iteractive configuration expansion
CI (ICE-CI) can be selected as CIStep in CASSCF or called directly using the ICE block as described in
9.15. The ICE-CI offers a more traditional approach to get approximate full CI results. These options and
modules have their own sections in the manual with a detailed documentation on their usage and the available
properties.
Moreover, ORCA interfaces a determinant-based FCI toy-code, which is described here. Besides energies, the
program can compute dipole moments, g-tensors, hyperfine and quadrupole splittings in the framework of
linear response theory. For response properties, it is assumed that the basis set does not depend on the given
perturbation. To call the program prepare a simple SCF input and add the keyword !FCI on top.
!def2-svp extremescf nofrozencore FCI
*xyz 0 2
Li 0 0 0
By default, energies and properties are computed for the ground state. For the g-tensor computation, the
origin is chosen as the center of electronic charge, while for dipoles it is the center of mass. The default
settings may be edited in the FCI block:
ETol 1e-10 # Energy convergence threshold

RTol 1e-8 # Residual convergence threshold
RespTol 1e-8 # Convergence threshold for the response equations
NRoots 1 # Number of states to be computed

DoGTensor true
SOCG true # SOC contribution to the g-tensor
DoHFC true
SOCHFC false # SOC contribution to the HFS
DoQuad true
DoDipole true
PrintLevel 3 # PrintLevel
ORIGIN -3 # (see EPRMR-module)
The input and output resembles the ORCA standard format from the EPRNMR module and should thus be
familiar. However, it is much more limited in the number of available options and the size of molecules that
can be treated. Currently it can only be used in serial mode and in combination with the nofrozencore
option.
8.1.10 Scalar Relativistic SCF
Scalar relativistic all-electron calculations can be performed with a variety of relativistic approximations.
However, these need to be combined with a suitable basis set since relativity changes the shapes of orbitals
considerably. We have defined scalar relativistic contracted versions of the QZV, TZV and SV basis sets up
to Hg for HF and DFT computations (but not yet for RI-MP2). They are requested by putting DKH- or
ZORA- in front of the usual basis set name. For other basis sets you have to take care of the recontraction
yourself but note that this is an expert issue. All scalar relativistic models can be used for geometry
optimization as well.
CAUTION:
For geometry optimizations we apply a one-center relativistic correction. This slightly

changes the energies so DO NOT MIX single-point calculations without the one-center
approximation with geometry optimization energies that DO make use of this feature.
The impact of the one-center approximation on the geometries is very small.
8.1.10.1 Douglas-Kroll-Hess
ORCA has implemented the standard second-order DKH procedure that is normally satisfactory for all
intents and purposes. The scalar relativistic DKH treatment is compatible with any of the SCF methods and
will also be transferred over to the correlation treatments.
We rather strongly recommend the use of the SV, TZV and QZV basis sets with or without def2 and
appropriate polarization functions. For these basis sets we have developed segmented relativistic all electron
basis sets for almost the entire periodic table. The basis sets are tested and perform very well in an acceptably
economic fashion.
The use of the code is very simple:
! UHF DKH-TZV DKH
NOTE: You should have the basis set and ZORA or DKH commands in the same input line!
8.1.10.2 ZORA and IORA
In addition to the DKH method the 0th order regular approximation (ZORA; pioneered by van Lenthe
et al., see Ref. [122] and many follow up papers by the Amsterdam group) is implemented into ORCA in
an approximate way (section 9.14) which facilitates the calculation of analytical gradients. Our ZORA
implementation essentially follows van W ullen [123] and solves the ZORA equations with a suitable model
potential which is derived from accurate atomic ZORA calculations. At this point the elements up to atomic
number 86 are available with more to come. The ZORA method is highly dependent on numerical integration
and it is very important to pay attention to the subject of radial integration accuracy (vide infra)! If the
relevant precautions are taken, the use of the ZORA or IORA methods is as easy as in the DKH case. For
example:
! UHF ZORA-TZVP ZORA

# for more detail use
%rel method ZORA # or IORA
modelpot 1,1,1,1
modeldens rhoZORA
velit 137.0359895 # speed of light used
end
ATTENTION
The scalar relativistic module has many options which allow you to considerably finetune the calculations.
Details are in section 9.14.
The scalar relativistic treatment requires flexible basis sets, in particular in the core region. Only SV,
TZV and QZV basis sets have been recontracted in the ZORA and DKH models (and the attached
polarization functions of course). Alternatively, one choice that you have is to uncontract your basis
set using the Decontract keyword but it is likely that you also need additional steep primitives.
Scalar relativistic calculations may need larger integration grids in the radial part. Consider to use
a higher IntAcc parameter or at least to increase the radial integration accuracy around the heavy
atoms using SpecialGridAtoms and SpecialGridIntAcc.
The calculation of properties in relativistic treatments is not straightforward since the influence of the
small component in the Dirac equation is neglected in the calcuation of expectation values. ORCA
takes these picture change effects to some extent into account. Please refer to individual sections.
8.1.11 Efficient Calculations with Atomic Natural Orbitals
Atomic natural orbitals are a special class of basis sets. They are represented by the orthonormal set of
orbitals that diagonalizes a spherically symmetric, correlated atomic density. The idea is to put as much
information as possible into each basis functions such that one obtains the best possible result with the given
number of basis functions. This is particularly important for correlated calculations where the number of
primitives is less an issue than the number of basis functions.
Usually, ANO basis sets are generally contracted which means that for any given angular momentum all
primitives contribute to all basis functions. Since the concept of ANOs only makes sense if the underlying set
of primitives is large, the calculations readily become very expensive unless special precaution is taken in the
integral evaluation algorithms. ORCA features special algorithms for ANO basis sets together with accurate
ANO basis sets for non-relativistic calculations. However, even then the integral evaluation is so expensive
that efficiency can only be realized if all integrals are stored on disk and are re-used as needed.
In the first implementation, the use of ANOs is restricted to the built-in ANO basis sets (ano-pVnZ, saug-
ano-pVnZ, aug-ano-pVnZ, n = D, T, Q, 5). These are built upon the cc-pV6Z primitives and hence, the
calculations take significant time.
NOTE:
There are no relativistic ANOs so far.
Geometry optimizations with ANOs are discouraged; they will be very inefficient.
The use of ANOs is recommended in the following way:
! ano-pVTZ Conv TightSCF CCSD(T)

%maxcore 2000
* int 0 1
C 0 0 0 0 0 0
O 1 0 0 1.2 0 0
H 1 2 0 1.1 120 0
H 1 2 3 1.1 120 180
*
This yieds:
ano-pVTZ:
E(SCF) = -113.920388785
E(corr)= -0.427730189
Compare to the cc-pVTZ value of:
cc-pVTZ:
E(SCF) = -113.911870901
E(corr)= -0.421354947
Thus, the ANO-based SCF energy is ca. 89 mEh lower and the correlation energy almost 2 mEh lower than
with the cc-basis set of the same size. Usually, the ANO results are much closer to the basis set limit than
the cc-results. Also, ANO values extrapolate very well (see section 8.1.3.5)
Importantly, the integrals are all stored in this job. Depending on your system and your patience, this
may be possible up to 300500 basis functions. The ORCA correlation modules have been rewritten such
that they deal efficiently with these stored integrals. Thus, we might as well have used ! MO-CCSD(T) or
! AO-CCSD(T), both of which would perform well.
Yet, the burden of generating and storing all four-index integrals quickly becomes rather heavy. Hence,
the combination of ANO basis sets with the RI-JK technique is particularly powerful and efficient. For
example:
! ano-pVTZ cc-pVTZ/JK RI-JK Conv TightSCF RI-CCSD(T)
For the SCF, this works very well and allows for much larger ANO based calculations to be done efficiently.
Also, RI-MP2 can be done very efficiently in this way. However, for higher order correlation methods such
as CCSD(T) the logical choice would be RI-CCSD(T) which is distinctly less efficient than the AO or MO
based CCSD(T) (roughly a factor of two slower). Hence, ORCA implements a hybrid method where the RI
approximation is used to generate all four index integrals. This is done via the RI-AO keyword:
! ano-pVTZ cc-pVTZ/JK RI-AO Conv TightSCF AO-CCSD(T)
In this case either AO-CCSD(T) or MO-CCSD(T) would both work well. This does not solve the storage
bottleneck with respect to the four index integrals of course. If this becomes a real issue, then RI-CCSD(T)
is mandatory. The error in the total energy is less than 0.1 mEh in the present example.
NOTE:
With conventional RI calculations the use of a second fit basis set is not possible and
inconsistent results will be obtained. Hence, stick to one auxiliary basis!
8.2 Geometry Optimizations, Surface Scans, Transition States, MECPs
The usage of analytic gradients is necessary for efficient geometry optimization. In ORCA4.0, the following
methods provide analytic first derivatives
Hartree-Fock (HF) and DFT (including the RI approximation)
MP2 (only canonical)
TD-DFT for excited states
CAS-SCF
When the analytic gradients are not available, it is possible to evaluate the first derivatives numerically by
finite displacements. This is available for all methods.
The coordinate system chosen for geometry optimization affects the convergence rate, with redundant internal
coordinates being usually the best choice.
The optimization of transition state (TS) require second derivative matrix (Hessian), implemented analytically
for HF, DFT and MP2. Additionally several approaches to construct an initial approximate Hessian for
TS optimization are available. Another interesting feature are MECP (Minimum Energy Crossing Point)
optimizations.
8.2.1 Geometry Optimizations
Optimizations are fairly easy as in the following example:
! RKS B3LYP/G SV(P) TightSCF Opt

* int 0 1
C 0 0 0 0.0000 0.000 0.00
O 1 0 0 1.2029 0.000 0.00
H 1 2 0 1.1075 122.016 0.00
H 1 2 3 1.1075 122.016 180.00
*
An optimization with the RI method (the BP functional is recommend) would simply look like:
! RKS BP RI SV(P) def2/J TightSCF Opt

* int 0 1
C 0 0 0 0.0000 0.000 0.00
O 1 0 0 1.2029 0.000 0.00
H 1 2 0 1.1075 122.016 0.00
H 1 2 3 1.1075 122.016 180.00
*
8.2 Geometry Optimizations, Surface Scans, Transition States, MECPs 137
An optimization of the first excited state of ethylene:
! BLYP SVP OPT
%tddft
IRoot 1
end
* xyz 0 1
C 0.000000 0.000000 0.666723
C 0.000000 0.000000 -0.666723
H 0.000000 -0.928802 1.141480
H -0.804366 -0.464401 -1.341480
H 0.000000 0.928802 1.241480
H 0.804366 0.464401 -1.241480
*
8.2.2 Numerical Gradients
If the analytic gradient is not available, the numerical gradient can simply be requested via:
! NumGrad
as in the following example:
!RHF CCSD(T) TZVPP

!Opt NumGrad
* int 0 1
C 0 0 0 0 0 0
O 1 0 0 1.2 0 0
H 1 2 0 1.1 120 0
H 1 2 3 1.1 120 180
*
NOTE
Be aware that the numerical gradient is quite expensive. The time for one gradient calculation is equal
to 6 (number of atoms) (time for one single point calculation).
The numerical gradient can be calculated in a multi-process run, using a maximum of three times the
number of atoms (see section 5.2).
More details on various options, geometry convergence criteria and the like are found in section 9.17.
8.2.3 Some Notes and Tricks
NOTE
TightSCF in the SCF part is set as default to avoid the buildup of too much numerical noise in the
gradients.
TIP
If you want to be on the safe side with DFT it is probably advisable to use the settings ! Grid4
NoFinalGrid although the defaults should also be o.k.
In rare cases the redundant internal coordinate optimization fails. In this case, you may try to use
COPT (optimization in Cartesian coordinates). If this optimization does not converge, you may try the
desperate choice to use ZOPT, GDIIS-COPT or GDIIS-ZOPT. This will likely take many more steps to
converge but should be stable.
For optimizations in Cartesian coordinates the initial guess Hessian is constructed in internal coordinates
and thus these optimizations should converge only slightly slower than those in internal coordinates.
Nevertheless, if you observe a slow convergence behaviour, it may be a good idea to compute a Hessian
initially (perhaps at a lower level of theory) and use InHess read in order to improve convergence.
At the beginning of a TS optimization more information on the curvature of the PES is needed than
a model Hessian can give. The best choice is analytic Hessian, avialable for HF, DFT and MP2. In
other cases (e.g. CAS-SCF), the numerical evaluation is necessary. Nevertheless you do not need to
calculate the full Hessian when starting such a calculation. With ORCA we have good experience
with approximations to the exact Hessian. Here it is recommended to either directly combine the
TS optimization with the results of a relaxed surface scan or to use the Hybrid Hessian as the initial
Hessian, depending on the nature of the TS mode. Note that these approximate Hessians do never
replace the exact Hessian at the end of the optimization, which is always needed to verify the minimum
or first order saddle point nature of the obtained structure.
8.2.4 Initial Hessian for Minimization
The convergence of a geometry optimization crucially depends on the quality of the initial Hessian. In the
simplest case it is taken as a unit matrix (in redundant internal coordinates we use 0.5 for bonds, 0.2 for angles
and 0.1 for dihedrals and improper torsions). However, simple model force-fields like the ones proposed by
Schlegel, Lindh or Almlof are available and lead to much better convergence. The different guess Hessians can
be set via the InHess option which can be either unit, Almloef, Lindh or Schlegel in redundant internal
coordinates. Since version 2.5.30, these model force-fields (built up in internal coordinates) can also be used
in optimizations in Cartesian coordinates.
For minimizations we recommend the Almloef Hessian, which is the default for minimizations. The Lindh
and Schlegel Hessian yield a similar convergence behaviour. Of course the best Hessian is the exact one.
Read may be used to input an exact Hessian or one that has been calculated at a lower level of theory (or a
faster level of theory). From version 2.5.30 on this option is also available in redundant internal coordinates.
But we have to point out that the use of the exact Hessian as initial one is only of little help, since in these
cases the convergence is usually only slightly faster, while at the same time much more time is spent in the
calculation of the initial Hessian.
To sum it up: we advise to use one of the simple model force-fields for minimizations.
8.2.5 Coordinate Systems for Optimizations
The coordinate system for the optimization can be chosen by the coordsys variable that can be set to
cartesian, or redundant new, or redundant, or redundant old) %geom block. The default is the redundant
internal coordinate system. (redundant old is the coordinate set that was used as default redundant internal
coordinates before version 2.4.30). If the optimization with redundant fails, redundant old can still be used
as an alternative, and in cases where the internal coordinate systems lead to problems, you can still try
cartesian. If the optimization is then carried out in Cartesian displacement coordinates with a simple model
force-field Hessian, the convergence will be only slightly slower. With a unit matrix initial Hessian very slow
convergence will result.
A job that starts from a semi-empirical Hessian is shown below:
# ---------------------------------------------------
# First calculate the frequencies at the input
# geometry which is deliberately chosen poorly
# ---------------------------------------------------
! RHF AM1 NumFreq
%base "FirstJob"
* int 0 1
C 0 0 0 0 0 0
O 1 0 0 1.3 0 0
H 1 2 0 1.1 110 0
H 1 2 3 1.1 110 180
*
$new_job
# --------------------------------------------------------
# Now the real job
# --------------------------------------------------------
! B3LYP SVP TightSCF PModel
! Opt
%base "SecondJob"
%geom GDIISMaxEq 20
UseGDIIS false
InHess Read
InHessName "FirstJob.hess"
# this file must be either a .hess file from a
# frequency run or a .opt file left over from a
# previous geometry optimization
end
* int 0 1
C 0 0 0 0 0 0
O 1 0 0 1.3 0 0
H 1 2 0 1.1 110 0
H 1 2 3 1.1 110 180
*
NOTE:
The guess PModel is chosen for the second job since DFT or HF calculations cannot be started from
semi-empirical orbitals
GDIIS has been turned off and the number of gradients used in the quasi-Newton method has been
enhanced. This is advisable if a good starting Hessian is available.
TIP
For transition metal complexes MNDO, AM1 or PM3 Hessians are not available. You can use ZINDO/1
or NDDO/1 Hessians instead. They are of lower quality than MNDO, AM1 or PM3 for organic
molecules but they are still far better than the standard unit matrix choice.
If the quality of the initial semi-empirical Hessian is not sufficient you may use a quick RI-DFT job
(e.g. BP Def-1 NoFinalGrid)
In semi-empirical geometry optimizations on larger molecules or in general when the molecules become
larger the redundant internal space may become large and the relaxation step may take a significant
fraction of the total computing time. In this case you may revert to delocalized internal coordinates
which only keep the non-redundant degrees of freedom and leads to more efficient update steps. It is
slightly less robust though.
8.2.6 Constrained Optimizations
You can perform constrained optimizations which can, at times, be extremely helpful. This works as shown
in the following example:

%geom Constraints
{ B 0 1 1.25 C }
{ A 2 0 3 120.0 C }
end
end
* int 0 1
C 0 0 0 0.0000 0.000 0.00
O 1 0 0 1.2500 0.000 0.00
H 1 2 0 1.1075 122.016 0.00

H 1 2 3 1.1075 122.016 180.00
*
Constraining bond distances : { B N1 N2 value C }

Constraining bond angles : { A N1 N2 N1 value C }
Constraining dihedral angles : { D N1 N2 N3 N4 value C }
Constraining cartesian coordinates : { C N1 C }
NOTE:
Like for normal optimizations you can use numerical gradients (see 8.2.2.) for constrained optimizations.
In this case the numerical gradient will be evaluated only for non-constrained coordinates, saving a lot
of computational effort, if a large part of the structure is constrained.
value in the constraint input is optional. If you do not give a value, the present value in the structure
is constrained. For cartesian constraints you cant give a value, but always the initial position is
constrained.
It is recommended to use a value not too far away from your initial structure.
It is possible to constrain whole sets of coordinates:
all bond lengths where N1 is involved : { B N1 * C}

all bond lengths : { B * * C}
all bond angles where N2 is the central atom: { A * N2 * C }
all bond angles : { A * * * C }
all dihedral angles with central bond N2-N3 : { D * N2 N3 * C }
all dihedral angles : { D * * * * C }
If there are only a few coordinates that have to be optimized you can use the invertConstraints option:
%geom Constraints
{ B 0 1 C }
end
invertConstraints true # only the C-O distance is optimized
# does not affect Cartesian coordinates
end
In some cases it is advantageous to optimize only the positions of the hydrogen atoms and let the
remaining molecule skeleton fixed:
%geom optimizehydrogens true

end
NOTE:
In the special case of a fragment optimization (see next point) the optimizehydrogens keyword does
not fix the heteroatoms, but ensures that all hydrogen positions are relaxed.
8.2.7 Constrained Optimizations for Molecular Clusters (Fragment Optimization)
If you want to study systems, which consist of several molecules (e.g. the active site of a protein) with
constraints, then you can either use cartesian constraints (see above) or use ORCAs fragment constraint
option. ORCA allows the user to define fragments in the system. For each fragment one can then choose
separately whether it should be optimized or constrained. Furthermore it is possible to choose fragment pairs
whose distance and orientation with respect to each other should be constrained. Here, the user can either
define the atoms which make up the connection between the fragments, or the program chooses the atom pair
automatically via a closest distance criterion. ORCA then chooses the respective constrained coordinates
automatically. An example for this procedure is shown below.
The coordinates are taken from a crystal structure [PDB-code 2FRJ]. In our gas phase model we choose only
a small part of the protein, which is important for its spectroscopic properties. Our selection consists of a
heme-group (fragment 1), important residues around the reaction site (lysine (fragment 2) and histidine
(fragment 3)), an important water molecule (fragment 4), the NO-ligand (fragment 5) and part of a histidine
(fragment 6) coordinated to the heme-iron. In this constrained optimization we want to maintain the position
of the heteroatoms of the heme group. Since the protein backbone is missing, we have to constrain the
orientation of lysine and histidine (fragments 2 and 3) side chains to the heme group. All other fragments
(the ones which are directly bound to the heme-iron and the water molecule) are fully optimized internally
and with respect to the other fragments. Since the crystal structure does not reliably resolve the hydrogen
positions, we relax also the hydrogen positions of the heme group.
# !! If you want to run this optimization: be aware

# !! that it will take some time!
! UKS BP86 RI SV(P) Opt TightSCF Grid4 NoFinalGrid
%geom
ConstrainFragments { 1 } end # constrain all internal
# coordinates of fragment 1
ConnectFragments
{1 2 C 12 28} # connect the fragments via the atom pair 12/28 and 15/28 and
{1 3 C 15 28} # constrain the internal coordinates connecting
# fragments 1/2 and 1/3
{1 5 O}
{1 6 O}
{2 4 O}
{3 4 O}
end
optimizeHydrogens true # do not constrain any hydrogen position
end
* xyz 1 2
Fe(1) -0.847213 -1.548312 -1.216237 newgto "TZVP" end
N(5) -0.712253 -2.291076 0.352054 newgto "TZVP" end
O(5) -0.521243 -3.342329 0.855804 newgto "TZVP" end
N(6) -0.953604 -0.686422 -3.215231 newgto "TZVP" end
N(3) -0.338154 -0.678533 3.030265 newgto "TZVP" end
N(3) -0.868050 0.768738 4.605152 newgto "TZVP" end
N(6) -1.770675 0.099480 -5.112455 newgto "TZVP" end
N(1) -2.216029 -0.133298 -0.614782 newgto "TZVP" end
N(1) -2.371465 -2.775999 -1.706931 newgto "TZVP" end
N(1) 0.489683 -2.865714 -1.944343 newgto "TZVP" end
N(1) 0.690468 -0.243375 -0.860813 newgto "TZVP" end
N(2) 1.284320 3.558259 6.254287
C(2) 5.049207 2.620412 6.377683
C(2) 3.776069 3.471320 6.499073
C(2) 2.526618 2.691959 6.084652
C(3) -0.599599 -0.564699 6.760567
C(3) -0.526122 -0.400630 5.274274
C(3) -0.194880 -1.277967 4.253789
C(3) -0.746348 0.566081 3.234394
C(6) 0.292699 0.510431 -6.539061
C(6) -0.388964 0.079551 -5.279555
C(6) 0.092848 -0.416283 -4.078708
C(6) -2.067764 -0.368729 -3.863111
C(1) -0.663232 1.693332 -0.100834
C(1) -4.293109 -1.414165 -0.956846
C(1) -1.066190 -4.647587 -2.644424
C(1) 2.597468 -1.667470 -1.451465
C(1) -1.953033 1.169088 -0.235289
C(1) -3.187993 1.886468 0.015415
C(1) -4.209406 0.988964 -0.187584
C(1) -3.589675 -0.259849 -0.590758
C(1) -3.721903 -2.580894 -1.476315
C(1) -4.480120 -3.742821 -1.900939
C(1) -3.573258 -4.645939 -2.395341
C(1) -2.264047 -4.035699 -2.263491
C(1) 0.211734 -4.103525 -2.488426
C(1) 1.439292 -4.787113 -2.850669
C(1) 2.470808 -3.954284 -2.499593
C(1) 1.869913 -2.761303 -1.932055
C(1) 2.037681 -0.489452 -0.943105
C(1) 2.779195 0.652885 -0.459645
C(1) 1.856237 1.597800 -0.084165

C(1) 0.535175 1.024425 -0.348298
O(4) -1.208602 2.657534 6.962748
H(3) -0.347830 -1.611062 7.033565
H(3) -1.627274 -0.387020 7.166806
H(3) 0.121698 0.079621 7.324626
H(3) 0.134234 -2.323398 4.336203
H(3) -1.286646 1.590976 5.066768
H(3) -0.990234 1.312025 2.466155
H(4) -2.043444 3.171674 7.047572
H(2) 1.364935 4.120133 7.126900
H(2) 0.354760 3.035674 6.348933
H(2) 1.194590 4.240746 5.475280
H(2) 2.545448 2.356268 5.027434
H(2) 2.371622 1.797317 6.723020
H(2) 3.874443 4.385720 5.867972
H(2) 3.657837 3.815973 7.554224
H(2) 5.217429 2.283681 5.331496
H(2) 5.001815 1.718797 7.026903
H(6) -3.086380 -0.461543 -3.469767
H(6) -2.456569 0.406212 -5.813597
H(6) 1.132150 -0.595619 -3.782287
H(6) 0.040799 1.559730 -6.816417
H(6) 0.026444 -0.139572 -7.404408
H(6) 1.392925 0.454387 -6.407850
H(1) 2.033677 2.608809 0.310182
H(1) 3.875944 0.716790 -0.424466
H(1) 3.695978 -1.736841 -1.485681
H(1) 3.551716 -4.118236 -2.608239
H(1) 1.487995 -5.784645 -3.308145
H(1) -1.133703 -5.654603 -3.084826
H(1) -3.758074 -5.644867 -2.813441
H(1) -5.572112 -3.838210 -1.826943
H(1) -0.580615 2.741869 0.231737
H(1) -3.255623 2.942818 0.312508
H(1) -5.292444 1.151326 -0.096157
H(1) -5.390011 -1.391441 -0.858996
H(4) -1.370815 1.780473 7.384747
H(2) 5.936602 3.211249 6.686961
*
NOTE:
You have to connect the fragments in such a way that the whole system is connected.
You can divide a molecule into several fragments.

Since the initial Hessian for the optimization is based upon the internal coordinates: Connect the
fragments in a way that their real interaction is reflected.
This option can be combined with the definition of constraints, scan coordinates and the optimizeHy-
drogens option (but: its meaning in this context is different to its meaning in a normal optimization
run, relatively straightforward see section 9.17).
Can be helpful in the location of complicated transition states (with relaxed surface scans).
8.2.8 Relaxed Surface Scans
A final thing that comes in really handy are relaxed surface scans, i.e. you can scan through one variable
while all others are relaxed. It works as shown in the following example:

%geom Scan
B 0 1 = 1.35, 1.10, 12 # C-O distance that will be scanned
end
end
* int 0 1
C 0 0 0 0.0000 0.000 0.00
O 1 0 0 1.3500 0.000 0.00
H 1 2 0 1.1075 122.016 0.00
H 1 2 3 1.1075 122.016 180.00
*
In the example above the value of the bond length between C and O will be changed in 12 equidistant steps
from 1.35 down to 1.10
Angstroms and at each point a constrained geometry optimization will be carried
out.
NOTE:
If you want to perform a geometry optimization at a series of values with non-equidistant steps you
can give this series in square brackets, [ ]. The general syntax is as follows:
B N1 N2 = initial-value, final-value, NSteps

or:
B N1 N2 [value1 value2 value3 ... valueN]
In addition to bond lengths you can also scan bond angles and dihedral angles:
B N1 N2 = ... # bond length

A N1 N2 N3 = ... # bond angle
D N1 N2 N3 N4 = ... # dihedral angle
TIP
As in constrained geometry optimization it is possible to start the relaxed surface scan with a different
scan parameter than the value present in your molecule. But keep in mind that this value should not
be too far away from your initial structure.
A more challenging example is shown below. Here, the H-atom abstraction step from CH4 to OH-radical is
computed with a relaxed surface scan (vide supra). The job was run as follows:
! UKS B3LYP SV(P) TightSCF Opt SlowConv

%geom scan B 1 0 = 2.0, 1.0, 15 end end
* int 0 2
C 0 0 0 0.000000 0.000 0.000
H 1 0 0 1.999962 0.000 0.000
H 1 2 0 1.095870 100.445 0.000
H 1 2 3 1.095971 90.180 119.467
H 1 2 3 1.095530 95.161 238.880
O 2 1 3 0.984205 164.404 27.073
H 6 2 1 0.972562 103.807 10.843
*
It is obvious that the reaction is exothermic and passes through an early transition state in which the
hydrogen jumps from the carbon to the oxygen. The structure at the maximum of the curve is probably a
very good guess for the true transition state that might be located by a transition state finder.
You will probably find that such relaxed surface scans are incredibly useful but also time consuming. Even
the simple job shown below required several hundred single point and gradient evaluations (convergence
problems appear for the SCF close to the transition state and for the geometry once the reaction partners
actually dissociate this is to be expected). Yet, when you search for a transition state or you want to get
insight into the shapes of the potential energy surfaces involved in a reaction it might be a good idea to
use this feature. One possibility to ease the burden somewhat is to perform the relaxed surface scan with a
fast method and a smaller basis set and then do single point calculations on all optimized geometries with
a larger basis set and/or higher level of theory. At least you can hope that this should give a reasonable
approximation to the desired surface at the higher level of theory this is the case if the geometries at the
lower level are reasonable.
8.2.9 Multiple XYZ File Scans
A different type of scan is implemented in ORCA in conjunction with relaxed surface scans. Such scans
produce a series of structures that are typically calculated using some ground state method. Afterwards
one may want to do additional or different calculations along the generated pathway such as excited state
calculations or special property calculations. In this instance, the multiple XYZ scan feature is useful. If
you request reading from a XYZ file via:
Figure 8.15: Relaxed surface scan for the H-atom abstraction from CH4 by OH-radical
(B3LYP/SV(P)).
* xyzfile Charge Multiplicity FileName
this file could contain a number of structures. The format of the file is:
Number of atoms M
Comment line
AtomName1 X Y Z
AtomName2 X Y Z
...
AtomNameM X Y Z
>
Number of atoms M
Comment line
AtomName1 X Y Z
...
Thus, the structures are simply of the standard XYZ format, separated by a > sign. After the last structure
no > should be given but a blank line instead. The program then automatically recognizes that a multiple
XYZ scan run is to be performed. Thus, single point calculations are performed on each structure in sequence
and the results are collected at the end of the run in the same kind of trajectory.dat files as produced from
trajectory calculations.
In order to aid in using this feature, the relaxed surface scans produce a file called MyJob.allxyz that is of
the correct format to be re-read in a subsequent run.
8.2.10 Transition States
8.2.10.1 Introduction to Transition State Searches
If you provide a good estimate for the structure of the transition state (TS) structure, then you can find the
respective transition state with the following keywords (in this example we take the structure with highest
energy of the above relaxed surface scan):
! UKS B3LYP SV(P) TightSCF SlowConv OptTS

# performs a TS optimization with the EF-algorithm
# Transition state: H-atom abstraction from CH4 to OH-radical
%geom
Calc_Hess true # calculation of the exact Hessian
# before the first optimization step
end
* int 0 2
C 0 0 0 0.000000 0.000 0.000
H 1 0 0 1.285714 0.000 0.000
H 1 2 0 1.100174 107.375 0.000
H 1 2 3 1.100975 103.353 119.612
H 1 2 3 1.100756 105.481 238.889
O 2 1 3 1.244156 169.257 17.024
H 6 2 1 0.980342 100.836 10.515
*
NOTE:
You need a good guess of the TS structure. Relaxed surface scans can help in almost all cases (see also
example above).
For TS optimization (in contrast to geometry optimization) an exact Hessian, a Hybrid Hessian or a
modification of selected second derivatives is necessary.
Analytic Hessian evaluation is available for HF and SCF methods, including the RI and RIJCOSX
approximations and canonical MP2.
Check the eigenmodes of the optimized structure for the eigenmode with a single imaginary frequency.
You can also visualize this eigenmode with orca pltvib (section 8.13.3.5) or any other visualization
program that reads ORCA output files.
If the Hessian is calculated during the TS optimization, it is stored as basename.001.hess, if it is

recalculated several times, then the subsequently calculated Hessians are stored as basename.002.hess,
basename.003.hess, . . .
If you are using the Hybrid Hessian, then you have to check carefully at the beginning of the TS
optimization (after the first three to five cycles) whether the algorithm is following the correct mode
(see TIP below). If this is not the case you can use the same Hybrid Hessian again via the inhess read
keyword and try to target a different mode (via the TS Mode keyword, see below).
In the example above the TS mode is of local nature. In such a case you can directly combine the relaxed
surface scan with the TS optimization with the
! ScanTS
command, as used in the following example:
! UKS B3LYP SV(P) TightSCF SlowConv

! ScanTS # perform a relaxed surface scan and TS optimization
# in one calculation
%geom scan B 1 0 = 2.0, 1.0, 15 end end
* int 0 2
C 0 0 0 0.000000 0.000 0.000
H 1 0 0 1.999962 0.000 0.000
H 1 2 0 1.095870 100.445 0.000
H 1 2 3 1.095971 90.180 119.467
H 1 2 3 1.095530 95.161 238.880
O 2 1 3 0.984205 164.404 27.073
H 6 2 1 0.972562 103.807 10.843
*
NOTE:
The algorithm performs the relaxed surface scan, aborts the Scan after the maximum is surmounted,
chooses the optimized structure with highest energy, calculates the second derivative of the scanned
coordinate and finally performs a TS optimization.
If you do not want the scan to be aborted after the highest point has been reached but be carried out
up to the last point, then you have to type:
%geom
fullScan true # do not abort the scan with !ScanTS
end
As transition state finder we implemented the quasi-Newton like hessian mode following algorithm. [124132]
This algorithm maximizes the energy with respect to one (usually the lowest) eigenmode and minimizes with
respect to the remaining 3N 7(6) eigenmodes of the Hessian.
TIP
You can check at an early stage if the optimization will lead to the correct transition state. After
the first optimization step you find the following output for the redundant internal coordinates:
---------------------------------------------------------------------------
Definition Value dE/dq Step New-Value frac.(TS mode)
----------------------------------------------------------------------------
1. B(H 1,C 0) 1.2857 0.013136 0.0286 1.3143 0.58
2. B(H 2,C 0) 1.1002 0.014201 -0.0220 1.0782
3. B(H 3,C 0) 1.1010 0.014753 -0.0230 1.0779
4. B(H 4,C 0) 1.1008 0.014842 -0.0229 1.0779
5. B(O 5,H 1) 1.2442 -0.015421 -0.0488 1.1954 0.80
6. B(H 6,O 5) 0.9803 0.025828 -0.0289 0.9514
7. A(H 1,C 0,H 2) 107.38 -0.001418 -0.88 106.49
8. A(H 1,C 0,H 4) 105.48 -0.002209 -0.46 105.02
9. A(H 1,C 0,H 3) 103.35 -0.003406 0.08 103.43
10. A(H 2,C 0,H 4) 113.30 0.001833 0.35 113.65
11. A(H 3,C 0,H 4) 113.38 0.002116 0.26 113.64
12. A(H 2,C 0,H 3) 112.95 0.001923 0.45 113.40
13. A(C 0,H 1,O 5) 169.26 -0.002089 4.30 173.56
14. A(H 1,O 5,H 6) 100.84 0.003097 -1.41 99.43
15. D(O 5,H 1,C 0,H 2) 17.02 0.000135 0.24 17.26
16. D(O 5,H 1,C 0,H 4) -104.09 -0.000100 0.52 -103.57
17. D(O 5,H 1,C 0,H 3) 136.64 0.000004 0.39 137.03
18. D(H 6,O 5,H 1,C 0) 10.52 0.000078 -0.72 9.79
----------------------------------------------------------------------------
Every Hessian eigenmode can be represented by a linear combination of the redundant internal coordinates.
In the last column of this list the internal coordinates, that represent a big part of the mode which is followed
uphill, are labelled. The numbers reflect their magnitude in the TS eigenvector (fraction of this internal
coordinate in the linear combination of the eigenvector of the TS mode). Thus at this point you can already
check whether your TS optimization is following the right mode (which is the case in our example, since we
are interested in the abstraction of H1 from C0 by O5.
If you want the algorithm to follow a different mode than the one with lowest eigenvalue, you can
either choose the number of the mode:
%geom
TS_Mode {M 1} # {M 1} mode with second lowest eigenvalue
end # (default: {M 0}, mode with lowest eigenvalue)
end
or you can give an internal coordinate that should be strongly involved in this mode:
%geom
TS_Mode {B 1 5} # bond between atoms 1 and 5,
end # you can also choose an angle: {A N1 N2 N1}
# or a dihedral: {D N1 N2 N3 N4}
end
8.2.10.2 Hessians for Transition State Calculations
For transition state (TS) optimization a simple initial Hessian, which is used for minimization, is not sufficient.
In a TS optimization we are looking for a first order saddle point, and thus for a point on the PES where the
curvature is negative in the direction of the TS mode (the TS mode is also called transition state vector, the
only eigenvector of the Hessian at the TS geometry with a negative eigenvalue). Starting from an initial
guess structure the algorithm used in the ORCA TS optimization has to climb uphill with respect to the TS
mode, which means that the starting structure has to be near the TS and the initial Hessian has to account
for the negative curvature of the PES at that point. The simple force-field Hessians cannot account for this,
since they only know harmonic potentials and thus positive curvature.
The most straightforward option in this case would be (after having looked for a promising initial guess
structure with the help of a relaxed surface scan) to calculate the exact Hessian before starting the TS
optimization. With this Hessian (depending on the quality of the initial guess structure) we know the TS
eigenvector with its negative eigenvalue and we have also calculated the exact force constants for all other
eigenmodes (which should have positive force constants). For the HF, DFT methods and MP2, the analytic
Hessian evaluation is available and is the best choice, for details see section Frequencies (8.3).
When only the gradients are available (most notably the CASSCF), the numerical calculation of the exact
Hessian is very time consuming, and one could ask if it is really necessary to calculate the full exact Hessian
since the only special thing (compared to the simple force-field Hessians) that we need is the TS mode with a
negative eigenvalue.
Here ORCA provides two different possibilities to speed up the Hessian calculation, depending on the nature
of the TS mode: the Hybrid Hessian and the calculation of the Hessian value of an internal coordinate. For
both possibilities the initial Hessian is based on a force-field Hessian and only parts of it are calculated
exactly. If the TS mode is of very local nature, which would be the case when e.g. cleaving or forming a bond,
then the exactly calculated part of the Hessian can be the second derivative of only one internal coordinate,
the one which is supposed to make up the TS mode (the formed or cleaved bond). If the TS mode is more
complicated and more delocalized, as e.g. in a concerted proton transfer reaction, then the hybrid Hessian,
a Hessian matrix in which the numerical second derivatives are calculated only for those atoms, which are
involved in the TS mode (for more details, see section 9.17), should be sufficient. If you are dealing with
more complicated cases where these two approaches do not succeed, then you still have the possibility to
start the TS optimization with a full exact Hessian.
Numerical Frequency calculations are quite expensive. You can first calculate the Hessian at a lower level of
theory or with a smaller basis set and use this Hessian as input for a subsequent TS optimization:
%geom inhess Read # this command comes with:

InHessName "yourHessian.hess" # filename of Hessian input file
end
Another possibility to save computational time is to calculate exact Hessian values only for those atoms which
are crucial for the TS optimization and to use approximate Hessian values for the rest. This option is very
useful for big systems, where only a small part of the molecule changes its geometry during the transition and
hence the information of the full exact Hessian is not necessary. With this option the coupling of the selected
atoms are calculated exactly and the remaining Hessian matrix is filled up with a model initial Hessian:
%geom Calc_Hess true

Hybrid_Hess [0 1 5 6] end # calculates a Hybrid Hessian with
end # exact calculation for atoms 0, 1, 5 and 6
For some molecules the PES near the TS can be very far from ideal for a Newton-Raphson step. In such a
case ORCA can recalculate the Hessian after a number of steps:
%geom
Recalc_Hess 5 # calculate the Hessian at the beginning
end # and recalculate it after 5,10,15,... steps
Another solution in that case is to switch on the trust radius update, which reduces the step size if the
Newton-Raphson steps behave unexpected and ensures bigger step size if the PES seems to be quite
quadratic:
%geom
Trust 0.3 # Trust <0 - use fixed trust radius (default: -0.3 au)
# Trust >0 - use trust radius update, i.e. 0.3 means:
# start with trust radius 0.3 and use trust radius update
end
8.2.10.3 Special Coordinates for Transition State Optimizations
If you look for a TS of a breaking bond the respective internal coordinate might not be included in
the list of redundant internal coordinates (but this would accelerate the convergence). In such a case
(and of course in others) you can add coordinates to or remove them from the set of autogenerated
redundant internal coordinates:
# add ( A ) or remove ( R ) internal coordinates

%geom
modify_internal
{ B 10 0 A } # add a bond between atoms 0 and 10
{ A 8 9 10 R } # remove the angle defined
# by atoms 8, 9 and 10
{ D 7 8 9 10 R } # remove the dihedral angle defined
end # by atoms 7, 8, 9 and 10
end
8.2.11 MECP Optimization
There are reactions where the analysis of only one spin state of a system is not sufficient, but where the
reactivity is determined by two or more different spin states (Two- or Multi-state reactivity). The analysis of
such reactions reveals that the different PESs cross each other while moving from one stationary point to the
other. In such a case you might want to use the ORCA optimizer to locate the point of lowest energy of the
crossing surfaces (called the minimum energy crossing point, MECP).
As an example for such an analysis we show the MECP optimization of the quartet and sextet state of
[FeO]+ .
!B3LYP TZVP Opt SurfCrossOpt SlowConv

%mecp
Mult 4
end
* xyz +1 6
Fe 0.000000 0.000000 1.000000
O 0.000000 0.000000 1.670000
*
For further options for the MECP calculation, see section 9.17.3.
TIP: You can often use a minimum or TS structure of one of the two spin states as initial guess for your
MECP-optimization. If this doesnt work, you might try a scan to get a better initial guess.
The results of the MECP optimization are given in the following output. The distance where both surfaces
cross is at 1.994
A. In this simple example there is only one degree of freedom and we can also locate the
MECP via a parameter scan. The results of the scan are given in Figure for comparison. Here we see that
the crossing occurs at a Fe-O-distance of around 2 A.
For systems with more than two atoms a scan is not sufficient any more and you have to use the MECP
optimization.
***********************HURRAY********************
*** THE OPTIMIZATION HAS CONVERGED ***
*************************************************
-------------------------------------------------------------------
-------------------------------------------------------------------
1. B(O 1,Fe 0) 1.9942 -0.000001 0.0000 1.9942
-------------------------------------------------------------------
*******************************************************
*** FINAL ENERGY EVALUATION AT THE STATIONARY POINT ***
Figure 8.16: Parameter scan for the quartet and sextet state of [FeO]+ (B3LYP/SV(P)).
*** (AFTER 8 CYCLES) ***

*******************************************************
------------------------------------- ----------------
Energy difference between both states -0.000002398
------------------------------------- ----------------
A more realistic example with more than one degree of freedom is the MECP optimization of a structure
along the reaction path of the CH3 O CH2 OH isomerization.
!B3LYP SV SurfCrossOpt SurfCrossNumFreq

%mecp Mult 1
end
*xyz 1 3
C 0.000000 0.000000 0.000000
H 0.000000 0.000000 1.300000
H 1.026719 0.000000 -0.363000
O -0.879955 0.000000 -1.088889
H -0.119662 -0.866667 0.961546
*
NOTE:
8.3 Vibrational Frequencies 155
To verify that a stationary point in a MECP optimization is a minimum, you have to use an adapted
frequency analysis, called by SurfCrossNumFreq (see section 9.17.3).
8.2.12 Constant External Force - Mechanochemistry
Constant external force can be applied on the molecule within the EFEI formalism [133] by pulling on the
two defined atoms. To apply the external force, use the POTENTIALS in the geom block. The potential
type is C for Constant force, indexes of two atoms (zero-based) and the value of force in nN.
! def2-svp OPT
%geom
POTENTIALS
{ C 2 3 4.0 }
end
end
* xyz 0 1
O 0.73020 -0.07940 -0.00000
O -0.73020 0.07940 -0.00000
H 1.21670 0.75630 0.00000
H -1.21670 -0.75630 0.00000
*
The results are seen in the output of the SCF procedure, where the total energy already contains the force
term.
----------------
TOTAL SCF ENERGY
----------------
Components:
Nuclear Repulsion : 36.90074715 Eh 1004.12038 eV
External potential : -0.25613618 Eh -6.96982 eV
Electronic Energy : -187.54166010 Eh -5103.26802 eV
8.3 Vibrational Frequencies
Vibrational frequency calculations are available through analytical differentiation of the SCF energy as well
as one- or two-sided numerical differentiation of analytical gradients, i.e. for Hartree-Fock and DFT models.
For methods without analytical gradient a numerically calculated gradient can be used (keyword NumGrad)
for numerical frequencies. Please note, that this will be a very time consuming calculation.
The use of vibrational frequency calculations is fairly simple:

# any Hartree-Fock or DFT model can be used here

! RKS BP def2-TZVP RI def2/J
# Tight SCF convergence and larger integration grids are advisable

# in order to minimize the numerical noise in the frequencies.
! TightSCF Grid5 NoFinalGrid
# perform a geometry optimization first

! Opt
# Run an analytical or numerical frequency calculation afterwards

! AnFreq # or just ! Freq
# numerical:
! NumFreq
# details of the analytical frequency calculation

%freq Hess2ElFlags 1,2,2,1 # use the RIJCOSX approximation
# (this is the new default for RIJCOSX Hessian)
end
# details of the numerical frequency calculation

%freq CentralDiff true # use central-differences (this is the default)
Increment 0.005 # increment in bohr for the
# differentiation (default 0.005)
end
! bohrs
* xyz 0 1
O -1.396288 -0.075107 0.052125
O 1.396289 -0.016261 -0.089970
H -1.775703 1.309756 -1.111179
H 1.775687 0.140443 1.711854
*
The user has full controll over approximations involved in the Hessian calculation. Hess2ElFlags(i1 , i2 ,
i3 , i4 ), where ik (ik = 0 to 2) adjust the use of the RI and COSX approximations. In case ik = 0 no
approximation in the 2-electron integrals is introduced. ik = 1 means the RI approximation in the Coulomb
part is in use. ik = 2 corresponds to COSX algorithm in the HF exchange. i1 corresponds to explicit Fock
matrix derivatives, i2 - Fock matrix depended on pseudo density, i3 - solution of the CP-SCF equations, i4 -
explicit integral second derivative.
In the introduced notation the RIJDX default flags are (1, 1, 1, 1), the RIJCOSX are (1, 2, 2, 1). Analytical
frequency calculations are also implemented at MP2 level. Please note, that the Hess2ElFlags are ignored
by the MP2 module. Furthermore, MP2 frequency calculations are very time-consuming and need a lot of
disk space.
At the end of the frequency job you get an output like this:
8.3 Vibrational Frequencies 157
-----------------------
VIBRATIONAL FREQUENCIES
-----------------------
0: 0.00 cm**-1
1: 0.00 cm**-1
2: 0.00 cm**-1
3: 0.00 cm**-1
4: 0.00 cm**-1
5: 0.00 cm**-1
6: 311.78 cm**-1
7: 887.65 cm**-1
8: 1225.38 cm**-1
9: 1394.81 cm**-1
10: 3624.88 cm**-1
11: 3635.73 cm**-1
This output consists of the calculated vibrational frequencies, the vibrational modes and the thermochemical
properties at 298.15 K. In the example above there are six frequencies which are identically zero. These
frequencies correspond to the rotations and translations of the molecule. They have been projected out of
the Hessian before the calculation of the frequencies and thus, the zero values do not tell you anything about
the quality of the Hessian that has been diagonalized. The reliability of the calculated frequencies has to
be judged by comparison of calculations with different convergence criteria, increments, integration grids
etc. The numerical error in the frequencies may reach 50 cm1 but should be considerably smaller in most
cases. Significant negative frequencies indicate saddle points of the energy hypersurface and prove that the
optimization has not resulted in an energy minimum.
Mass dependencies: Of course the calculated frequencies depend on the masses used for each atom.
While this can be influenced later through the orca vib routine (see Section 8.13.3.6 for more detail) and
individually for each atom in the geometry input, one might prefer using a set of precise atomic masses
rather than the set of atomic weights (which are set as default). This can be achieved through the !Mass2016
keyword, which triggers Orca to use those atomic masses representing either the most abundant isotope or
the most stable isotope (if all isotopes are unstable) of a certain element (e.g. the mass of 35 Cl for chlorine or
the mass of 98 Tc).
NOTE: The calculation of numerical frequencies puts rather high demands on both computer time and
accuracy. In order to get reliable frequencies make sure that:
Your SCF is tightly converged. A convergence accuracy of at least 107 Eh in the total energy and
106 in the density is desirable.
Grids of at least Grid4 or preferably larger are recommended.
The use of two-sided (i.e. central) differences increases the computation time by a factor of two but
gives more accurate and reliable results.
Small auxiliary basis sets like DGauss/J or DeMon/J may not result in fully converged frequencies
(up to 40 cm1 difference compared to frequencies calculated without RI). The def2/J universal
auxiliary basis sets of Weigend that are now the default in ORCA (or the SARC/J for scalar relativistic
calculations) are thought to give sufficiently reliable results.
Possibly, the convergence criteria of the geometry optimization need to be tightened in order to get
fully converged results.
If you can afford it, decrease the numerical increment to 0.001 Bohr or so. This puts even higher
demands on the convergence characteristics of the SCF calculation but should also give more accurate
numerical second derivatives. If the increment is too small or too high inaccurate results are expected.
The calculation of analytical frequencies is memory consuming. To control memory consumption the %maxcore
parameter must be set. For example %maxcore 8192 - use 8 Gb of memory per processor for the calculation.
The user should provide the value according to the computer available memory. The batching based on
%maxcore parameter will be introduced automatically to overcome probable memory shortage.
Numerical frequency calculations are restartable (but analytical frequency calculations are not). If the
numerical frequencies job died for one reason or another you can simply continue from where it stopped as in
the following example:
! RHF STO-3G
! NumFreq
%freq Restart true # restart an old calculation
# this requires a .hess file to be present
end
* int 0 1
C 0 0 0 0.0000 0 0
C 1 0 0 1.2160 0 0
H 1 2 0 1.083 180 0
H 2 1 3 1.083 180 0
*
NOTE
You must not change the level of theory, basis set or any other detail of the calculation. Any change
will produce an inconsistent, essentially meaningless Hessian.
The geometry at which the Hessian is calculated must be identical. If you followed a geometry
optimization by a frequency run then you must restart the numerical frequency calculation from the
optimized geometry.
Numerical frequencies can be performed in multi-process mode. Please see section 5.2 (Hints on the
use of parallel ORCA) for more information.
8.4 Reaction Energies from Compound Jobs 159
In multi-process mode the restart will take off from the locally calculated Hessians (.procmyid.hess,
e.g. .proc0.hess, .proc1.hess) where myid is the process id of some local process. Please make sure
that all these local Hessians get copied to your compute directory. If restart is set and no local files to
be found, ORCA will try to restart on the global Hessian file.
The Hessian file will contain a joblist, showing which displacements have been performed. You can
recalculate a single (or more) displacements by changing the 1 (done-) entry to 0 (to-be-done).
8.4 Reaction Energies from Compound Jobs
In order to make your life easier, we have implemented a mechanism that allows you to run several fragments
of a reaction in one input file and obtain the energy difference between reactants, intermediates and products
directly in the output file. The mechanism is such that you run a compound job and assign each fragment
that you wish to include in the energy profile a name:
%id "Educt1"
...
$new_job
%id "Educt2"
...
$new_job
%id "Product1"
...
In this example, there are three fragments. The first two are educts and the third one is a product. Thus
we look at a reaction A+BC. However, more complicated transformations may easily be specified in an
analogous way. The final energies are always referenced to the first fragment encountered in the file. It is
of course your responsibility that the fragments make sense and that the chemical reactions that your are
calculating are balanced with respect to nuclei and electrons.
Let us look at a very simple example, the reaction of H2 and CO to give H2 CO:
! B3LYP SV(P)
%id "eductA"
* xyz 0 1
H 0.000 0.000 0.000
H 0.75 0.000 0.000
*
$new_job
! B3LYP SV(P) PModel
%id "eductB"
* xyz 0 1
C 0.000 0.000 0.000
O 1.13 0.000 0.000

*
$new_job
! B3LYP SV(P) PModel
%id "product"
* int 0 1
C 0 0 0 0 0 0
O 1 0 0 1.2 0 0
H 1 2 0 1.1 120 0
H 1 2 3 1.1 120 180
*
NOTE:
You are not limited to educt and product. You might as well include intermediates, transition
states, minimum energy crossing points etc. in the list of fragments
If you do not have optimized geometries for each species you may want to run a geometry optimization at a
lower level and obtain energies at a higher level. This can be accomplished in an automatic way as follows:
# ---------------
# Now fragment 1: H2 opt at BP/SV(P) then single point CCSD(T)
# ---------------
! BP SV(P) TightSCF Opt PModel
%base "H2-BP"
* xyz 0 1
H 0.000 0.000 0.000
H 0.75 0.000 0.000
*
$new_job
! RHF CCSD(T) TZVPP
%id "educt"
%base "H2-CCSDT"
* xyzfile 0 1 H2-BP.XYZ
# ---------------
# Now fragment 2: CO opt at BP/SV(P) then single point CCSD(T)
# ---------------
$new_job
%base "CO-BP"
* xyz 0 1
C 0.000 0.000 0.000
O 1.13 0.000 0.000
*
$new_job
! RHF CCSD(T) TZVPP
8.5 Excited States with RPA, CIS, CIS(D), ROCIS and TD-DFT 161
%id "educt"
%base "CO-CCSDT"
* xyzfile 0 1 CO-BP.XYZ
# ---------------
# Now fragment 3: H2CO opt at BP/SV(P) then single point CCSD(T)
# ---------------
$new_job
%base "H2CO-BP"
* int 0 1
C 0 0 0 0 0 0
O 1 0 0 1.2 0 0
H 1 2 0 1.1 120 0
H 1 2 3 1.1 120 180
*
$new_job
! RHF CCSD(T) TZVPP
%id "product"
%base "H2CO-CCSDT"
* xyzfile 0 1 H2CO-BP.XYZ
NOTE:10
For each new fragment, the guess has to be explicitly specified since otherwise the program will try to
read it from the previous job and this is not a good idea since the number and identity of atoms does
not match
Note that %id is only given for the CCSD(T) jobs. This ensures that only these energies will occur
later in the calculation of the reaction energies.
8.5 Excited States with RPA, CIS, CIS(D), ROCIS and TD-DFT
ORCA features a module to perform TD-DFT, single-excitation CI (CIS) and RPA. The module works
with either closed-shell (RHF or RKS) or unrestricted (UHF or UKS) reference wavefunctions. For DFT
models the module automatically chooses TD-DFT and for HF wavefunctions the CIS model. If the RI
approximation is used in the SCF part it will also be used in the excited states calculation. A detailed
documentation is provided in section 9.18.
10
This job produces a CCSD(T) reaction energy not corrected for ZPE of 3.7 kcal/mol; B3LYP gets 11.2
kcal/mol a surprisingly large deviation and B2PLYP predicts 6.1 kcal/mol, which is certainly in the ballpark.
8.5.1 General Use
In its simplest form it is only necessary to provide the number of roots sought:
! RKS B3LYP/G SV(P) TightSCF
%tddft nroots 8
maxdim 2 #Davidson expansion space = MaxDim * NRoots
triplets true
end
* int 0 1
C(1) 0 0 0 0.00 0.0 0.00
O(2) 1 0 0 1.20 0.0 0.00
H(3) 1 2 0 1.08 120 0.00
H(3) 1 2 3 1.08 120 180.00
*
The MaxDim parameter controls the maximum number of expansion vectors in the iterative solution of the
CI equations. The default is the smallest possible choice. The triplets parameter is only valid for closed-shell
references. If chosen as true the program will also determine the triplet excitation energies in addition to the
singlets.
8.5.2 Use of TD-DFT for the Calculation of X-ray Absorption Spectra
In principle X-ray absorption spectra are normal absorption spectra that are just taken in a special
high-energy wavelength range. Due to the high energy of the radiation employed (several thousand eV),
core-electrons rather than valence electrons are excited. This has two consequences: a) the method becomes
element specific because the core-level energies divide rather cleanly into regions that are specific for a given
element. b) the wavelength of the radiation is so short that higher-order terms in the expansion of the
light-matter interaction become important. Most noticeably, quadrupole intensity becomes important.
X-ray absorption spectra can be generally divided into three regions: a) the pre-edge that corresponds to
transitions of core electrons into low lying virtual orbitals that lead to bound states. b) the rising edge that
corresponds to excitations to high-lying states that are barely bound, and c) the extended X-ray absorption
fine structure region (EXAFS) that corresponds to electrons being ejected from the absorber atom and
scattered at neighbouring atoms.
With the simple TD-DFT calculations described here, one focuses the attention on the pre-edge region.
Neither the rising edge nor the EXAFS region are reasonably described with standard electronic structure
methods and no comparison should be attempted. In addition, these calculations are restricted to K-edges as
the calculation of L-edges is much more laborious and requires a detailed treatment of the core hole spin
orbit coupling.
It is clearly hopeless to try to calculate enough states to cover all transitions from the valence to the pre-edge
region. Hence, instead one hand-selects the appropriate donor core orbitals and only allows excitations out of
these orbitals into the entire virtual space. This approximation has been shown to be justified. [134] One
should distinguish two situations: First, the core orbital in question may be well isolated and unambiguously
defined. This is usually the case for metal 1s orbitals if there is only one metal of the given type in the
molecule. Secondly, there may be several atoms of the same kind in the molecule and their core orbitals
form the appropriate symmetry adapted linear combinations dictated by group theory. In this latter case
special treatment is necessary: The sudden approximation dictates that the excitations occurs from a local
core orbital. In previous versions of the program you had to manually localize the core holes. In the present
version there is an automatic procedure that is described below.
A typical example is TiCl4 . If we want to calculate the titanium K-edge, the following input is appropriate:
! BP86 ZORA ZORA-def2-TZVP(-f) SARC/J TightSCF

%maxcore 1024
%tddft OrbWin[0] = 0,0,-1,-1

NRoots 25
DoQuad true
end
* int 0 1
Ti 0 0 0 0 0 0
Cl 1 2 3 2.15 0 0
Cl 1 2 3 2.15 109.4712 0
Cl 1 2 3 2.15 109.4712 120
Cl 1 2 3 2.15 109.4712 240
*
NOTE:
The absolute transition energies from such calculations are off by a few hundred electron volts due
to the shortcomings of DFT. The shift is constant and very systematic for a given element. Hence,
calibration is possible and has been done for a number of edges already. Calibration depends on the
basis set!
Quadrupole contributions (and magnetic dipole contributions) have been invoked with DoQuad true,
which is essential for metal edges. For ligand edges, the contributions are much smaller.
OrbWin is used to select the single donor orbital (in this case the metal 1s). The LUMO (45) and
last orbital in the set (174) are selected automatically if -1 is given. This is different from previous
program versions where the numbers had to be given manually.
The output contains standard TD-DFT output but also:

------------------------------------------------------------------------------------------------------
COMBINED ELECTRIC DIPOLE + MAGNETIC DIPOLE + ELECTRIC QUADRUPOLE SPECTRUM (origin adjusted)
------------------------------------------------------------------------------------------------------
State Energy Wavelength D2 m2 Q2 D2+m2+Q2 D2/TOT m2/TOT Q2/TOT
(cm-1) (nm) (*1e6) (*1e6)
------------------------------------------------------------------------------------------------------
....
This section contains the relevant output since it combines electric dipole, electric quadrupole and magnetic
dipole transition intensities into the final spectrum. Importantly, there is a gauge issue with the quadrupole
intensity: the results depend on the where the origin is placed. We have proposed a minimization procedure
that guarantees the fastest possible convergence of the multipole expansion. [135]
The spectra are plotted by calling
orca_mapspc MyOutput.out ABSQ -eV -x04890 -x14915 -w1.3
Starting from ORCA version 4.0 one may obtain nearly origin independent intensities by evaluating in addition
to the electric dipole (D), electric quadrupole (Q) and magnetic dipole (m) intensities the corresponding electric
dipole - magnetic quadrupole (DM) and the electric dipole - electric octupole (DO) intensities. [136] [137]
----------------------------------------------------------------------------------------------
COMBINED ELECTRIC DIPOLE + MAGNETIC DIPOLE + ELECTRIC QUADRUPOLE SPECTRUM
(Origin Independent, Length Representation)
----------------------------------------------------------------------------------------------
State Energy Wavelength D2 m2 Q2 DM DO D2+m2+Q2+DM+DO ...
(cm-1) (nm) (*1e6) (*1e6) (*1e6) (*1e6) (*1e6)
----------------------------------------------------------------------------------------------
....
These spectra are plotted by calling
orca_mapspc MyOutput.out ABSOI -eV -x04890 -x14915 -w1.3
Approximate origin independence is achieved by using the length approximation for distances from the
excited atom up to about 5 Angstrom.
Negative intensities may be found which can be partly cured by using larger basis sets.
In the cases where negative intensities appear in the spectrum it is presently recommended to use the
origin adjusted method
Now, let us turn to the Cl K-edge. Looking at the output of the first calculation, we have:
----------------
ORBITAL ENERGIES
----------------
NO OCC E(Eh) E(eV)

0 2.0000 -180.182806 -4903.0234
1 2.0000 -101.517692 -2762.4368
2 2.0000 -101.517681 -2762.4365
3 2.0000 -101.517442 -2762.4300
4 2.0000 -101.517442 -2762.4300
5 2.0000 -19.814042 -539.1675
6 2.0000 -16.394212 -446.1092
7 2.0000 -16.394210 -446.1091
8 2.0000 -16.394203 -446.1089
9 2.0000 -9.281040 -252.5499
10 2.0000 -9.281011 -252.5491
11 2.0000 -9.280892 -252.5459
12 2.0000 -9.280892 -252.5459
18 2.0000 -7.034911 -191.4297
....
And looking at the energy range or the orbital composition, we find that orbitals 1 through 4 are Cl 1s-orbitals.
They all have the same energy since they are essentially non-interacting. Hence, we can localize them without
invalidating the calculation. To this end, you can invoke the automatic localization for XAS which modifies
the input to:
! BP86 ZORA ZORA-def2-TZVP(-f) SARC/J TightSCF

%maxcore 1024
%tddft XASLoc[0] = 1,4

OrbWin[0] = 1,1,-1,-1
NRoots 25
DoQuad true
end
* int 0 1
Ti 0 0 0 0 0 0
Cl 1 2 3 2.15 0 0
Cl 1 2 3 2.15 109.4712 0
Cl 1 2 3 2.15 109.4712 120
Cl 1 2 3 2.15 109.4712 240
*
This localizes the orbitals 1 through 4 of operator 0 (the closed-shell) and then allows excitations
(arbitrarily) from core hole 1 only. You could choose any of the three other localized 1s orbitals instead
without changing the result. You could even do all four core holes simultaneously (they produce
identical spectra) in which case you have the entire ligand K-edge intensity and not just the one
normalized to a single chlorine (this would be achieved with OrbWin[0] = 1,4,-1,-1).
If you have a spin unrestricted calculation, you need to give the same XASLoc and OrbWin information
for the spin-down orbitals as well.
Quite nice results have been obtained for a number of systems in this way. [138]
8.5.3 Excited State Geometry Optimization
For RPA, CIS, TDA and TD-DFT the program can calculate analytic gradients. With the help of the IRoot
keyword, a given state can be selected for geometry optimization. Note however, that if two states cross
during the optimization it may fail to converge or fail to converge to the desired excited state!
! RHF SVP TightSCF Opt
%tddft NRoots 1
IRoot 1
end
* int 0 1
C(1) 0 0 0 0.00 0.0 0.00
O(2) 1 0 0 1.20 0.0 0.00
H(3) 1 2 0 1.08 120 0.00
H(3) 1 2 3 1.08 120 180.00
*
Note that this example converges to a saddle point as can be verified through a numerical frequency calculation
(which is also possible with the methods mentioned above). The excited state relaxed density matrix is
available from such gradient runs (MyJob.cisp when using the KeepDens keyword) and can be used for
various types of analysis. Note that the frozen core option is available starting from version 2.8.0.
8.5.4 Doubles Correction
For CIS (and also for perturbatively corrected double-xhybrid functionals) the program can calculate a
doubles correction to the singles-only excited states. The theory is due to Head-Gordon and co-workers.
%cis dcorr n # n=1,2,3,4 are four different algorithms that

# lead to (essentially) the same result but differ
# in the way the rate-limiting steps are handled
NOTE:
CIS(D) is often a quite big improvement over CIS.

The cost of the (D) correction is O(N5 ) and therefore comparable to RI-MP2. Since there are quite a
few things more to be done for (D) compared to RI-MP2, expect the calculations to take longer. In
the most elementary implementation the cost is about two times the time for RI-MP2 for each root.
The (D) correction is compatible with the philosophy of the double-hybrid density functionals and
should be used it these functionals are combined with TD-DFT. The program takes this as the default
but will not enforce it.
In our implementation it is only implemented together with the RI approximation and therefore you
need to supply an appropriate (/C) fitting basis.
The program will automatically put the RI-MP2 module into operation together with the (D) correction.
This will result in the necessary integrals becoming available to the CIS module.
8.5.5 Restricted Open-shell CIS
In addition to the CIS/TD-DFT description of excited states, ORCA features the orca rocis module to
perform configuration interaction with single excitations calculations using a restricted open-shell Hartree
Fock (ROHF) reference. It can be used to calculate excitation energies, absorption intensities and CD
intensities. In general, ROCIS calculations work on restricted open-shell HF reference functions but in this
implementation it is possible to enter the calculations with RHF (only for closed-shell molecules) or UHF
reference functions as well. If the calculation starts with an UHF/UKS calculation, it will automatically
produce the quasi-restricted orbitals which will then be used for the subsequent ROCIS calculations. Note
that if the reference function is a RHF/RKS function the method produces the CIS results. The module is
invoked by providing the number of roots sought in the %rocis block of the input file:
! SVP TightSCF
%rocis NRoots 2
MaxDim 5 #Davidson expansion space = MaxDim * NRoots
end
* xyz -2 2
Cu 0.00 0.00 0.00
Cl 2.25 0.00 0.00
Cl -2.25 0.00 0.00
Cl 0.00 2.25 0.00
Cl 0.00 -2.25 0.00
*
In this example the MaxDim parameter is given in addition to the number of roots to be calculated. It controls
the maximum dimension of the expansion space in the Davidson procedure that is used to solve the CI
problem.
The use of ROCIS is explained in greater detail in section 9.19.

8.6 SCF Stability Analysis
The SCF stability will give an indication whether the SCF solution is at a local minimum or a saddle
point. [139, 140] It is available for RHF/RKS and UHF/UKS. In the latter case, the SCF is restarted by
default using new unrestricted start orbitals if an instability was detected. For a demonstration, consider the
following input:
! BHLYP def2-SVP NORI
%scf
guess hcore
HFTyp UHF
STABPerform true
end
* xyz 0 1
h 0.0 0.0 0.0
h 0.0 0.0 1.4
*
The HCORE guess leads to a symmetric/restricted guess, which does not yield the unrestricted solution.
The same is often true for other guess options. For more details on the stability analysis, see section 9.7.
8.7 Excited States with EOM-CCSD
The methods described in the previous section are all based on the single excitation framework. For a
more accurate treatment, double excitations should also be considered. The equation of motion (EOM)
CCSD method (and the closely related family of linear response CC methods) provides an accurate way of
describing excited, ionized and electron attached states based on singles and doubles excitations within the
coupled-cluster framework. In this chapter, the typical usage of the EOM-CCSD routine will be described,
along with a short list of its present capabilities. A detailed description will be given in 9.20.
8.7.1 General Use
The simplest way to perform an EOM calculation is via the usage of the EOM-CCSD keyword, together with
the specification of the desired number of roots:
! RHF EOM-CCSD cc-pVDZ TightSCF
%mdci
nroots 9
8.7 Excited States with EOM-CCSD 169
end
*xyz 0 1
C 0.016227 -0.000000 0.000000
O 1.236847 0.000000 -0.000000
H -0.576537 0.951580 -0.000000
H -0.576537 -0.951580 -0.000000
*
The above input will call the EOM routine with default settings. The main output is a list of excitation
energies, augmented with some further state specific data. For the above input, the following output is
obtained:
----------------------
EOM-CCSD RESULTS (RHS)
----------------------
IROOT= 1: 0.147818 au 4.022 eV 32442.3 cm**-1

Amplitude Excitation
-0.107929 4 -> 8
0.665498 7 -> 8
0.104696 7 -> 8 6 -> 8
Ground state amplitude: 0.000000
IROOT= 2: 0.314137 au 8.548 eV 68945.2 cm**-1
-0.671249 7 -> 9
IROOT= 3: 0.343838 au 9.356 eV 75463.7 cm**-1
-0.670757 5 -> 8
-0.112213 6 -> 8 5 -> 8
IROOT= 4: 0.364197 au 9.910 eV 79931.9 cm**-1
-0.102822 4 -> 10
-0.484680 6 -> 8
0.438242 7 -> 10
-0.167485 6 -> 8 6 -> 8
Ground state amplitude: -0.021122
IROOT= 5: 0.389395 au 10.596 eV 85462.3 cm**-1
0.646790 4 -> 8
0.122394 7 -> 8
-0.171550 7 -> 8 6 -> 8
IROOT= 6: 0.414589 au 11.282 eV 90991.8 cm**-1
-0.378598 6 -> 8
-0.537683 7 -> 10
-0.124965 6 -> 8 6 -> 8
IROOT= 7: 0.423876 au 11.534 eV 93030.0 cm**-1
0.673885 7 -> 11
IROOT= 8: 0.444228 au 12.088 eV 97496.7 cm**-1
-0.664844 6 -> 9
-0.130443 6 -> 9 6 -> 8
IROOT= 9: 0.510518 au 13.892 eV 112045.8 cm**-1
0.665772 6 -> 10
0.114314 6 -> 15
0.125093 6 -> 10 6 -> 8
The IP and EA versions can be called using the keywords IP-EOM-CCSD and EA-EOM-CCSD respectively.
8.7.2 Capabilities
At present, the EOM routine is able to perform excited, ionized and electron attached state calculations,
for closed-shell systems, using an RHF reference function. It can be used for serial and parallel calculations.
A lower scaling version can be invoked by setting the CCSD2 keyword to true in the %mdci section. For
the time being, the most useful information provided is the list of the excitation energies, the ionization
potentials or the electron affinities. The ground to excited state transition moments are also available for
EE-EOM-CCSD.
A UHF EE-EOM-CCSD code is also available for the calculation of excitation energies in open-shell systems.
The usage is the same as the RHF implementation described above, once the RHF keyword has been replaced
by UHF. The IP- and EA- variants are not yet available in the current release. Also, the UHF EE-EOM-CCSD
implementation, in its current form, can only be used for serial calculations.
8.8 Excited States with STEOM-CCSD
The STEOM-CCSD method provides an efficient way to calculate excitation energies, with an accuracy
comparable to the EOM-CCSD approach, at a nominal cost. A detailed description will be given in 9.21.
8.8.1 General Use
The simplest way to perform a STEOM calculation is via the usage of the STEOM-CCSD keyword, together
with the specification of the desired number of roots:
8.8 Excited States with STEOM-CCSD 171
! RHF STEOM-CCSD cc-pVDZ TightSCF
%mdci
nroots 9
Dodbfilter true
end
*xyz 0 1
C 0.016227 -0.000000 0.000000
O 1.236847 0.000000 -0.000000
H -0.576537 0.951580 -0.000000
H -0.576537 -0.951580 -0.000000
*
The above input will call the STEOM routine with default settings. The main output is a list of excitation
energies, augmented with some further state specific data. The STEOMCC approach in ORCAuses state-
averaged CIS natural transition orbitals(NTO) for the selection of the active space. For the above input, the
following output is obtained:
------------------
STEOM-CCSD RESULTS
------------------
IROOT= 1: 0.146552 au 3.988 eV 32164.5 cm**-1

0.196225 4 -> 8
-0.979974 7 -> 8
Amplitude Excitation in Canonical Basis

-0.153212 4 -> 8
0.977931 7 -> 8
-0.121980 7 -> 13
IROOT= 2: 0.308608 au 8.398 eV 67731.7 cm**-1

-0.141414 4 -> 9
0.988498 7 -> 9

-0.989700 7 -> 9
IROOT= 3: 0.336979 au 9.170 eV 73958.3 cm**-1

-0.994070 5 -> 8

0.983934 5 -> 8
-0.137018 5 -> 13
IROOT= 4: 0.362974 au 9.877 eV 79663.6 cm**-1

0.177265 4 -> 10
0.825223 6 -> 8
-0.500412 7 -> 10
-0.118642 7 -> 12

-0.152751 4 -> 10
-0.821991 6 -> 8
0.506004 7 -> 10
IROOT= 5: 0.402096 au 10.942 eV 88249.9 cm**-1

0.100684 5 -> 11
0.617781 6 -> 8
0.761064 7 -> 10

-0.612814 6 -> 8
-0.754151 7 -> 10
IROOT= 6: 0.421001 au 11.456 eV 92399.1 cm**-1

-0.165095 4 -> 11
0.983905 7 -> 11

0.121348 4 -> 11
-0.983982 7 -> 11
IROOT= 7: 0.445178 au 12.114 eV 97705.3 cm**-1

0.995471 6 -> 9

-0.989647 6 -> 9
IROOT= 8: 0.462852 au 12.595 eV 101584.3 cm**-1

-0.985707 4 -> 8
-0.130220 6 -> 10

0.975461 4 -> 8
-0.147945 4 -> 13
0.128680 6 -> 10
IROOT= 9: 0.512757 au 13.953 eV 112537.1 cm**-1

0.121760 4 -> 8
-0.989185 6 -> 10

-0.121079 4 -> 8
0.979589 6 -> 10
-0.154643 6 -> 15
8.9 Excited States with PNO based coupled cluster methods 173
The first set of excitation amplitudes, printed for each root, have been calculated in the CIS NTO (Natural
Transition Orbitals) basis. The second set of amplitudes have been evaluated in the RHF canonical basis.
8.8.2 Capabilities
At present, the STEOM routine is able to calculate excitation energies, for closed-shell systems, using an
RHF reference function. It can be used for both serial and parallel calculations. A lower scaling version can
be invoked by setting the CCSD2 keyword to true in the %mdci section. The transition moments can also be
obtained.
8.9 Excited States with PNO based coupled cluster methods
The methods described in the previous section are performed over a canonical CCSD or MP2 ground state.
The use of canonical CCSD amplitudes restricts the use of EOM-CC and STEOM-CC methods to small
molecules. The use of MP2 amplitudes is possible (e.g. the EOM-CCSD(2) or STEOM-CCSD(2) approaches),
but it seriously compromises the accuracy of the method.
The bt-PNO-EOM-CCSD methods gives an economical compromise between accuracy and computational
cost by replacing the most expensive ground state CCSD calculation with a DLPNO based CCSD calculation.
The typical deviation of the results from the canonical EOM-CCSD results is around 0.01 eV. A detailed
description will be given in 9.22.
8.9.1 General Use
The simplest way to perform a PNO based EOM calculation is via the usage of the bt-PNO-EOM-CCSD
keyword, together with the specification of the desired number of roots. The specification of an auxilary basis
set is also required, just as for ground state DLPNO-CCSD calculations.
! RHF bt-PNO-EOM-CCSD cc-pVDZ cc-pVDZ/C cc-pVDZ/J TightSCF
%mdci
nroots 9
end
*xyz 0 1
C 0.016227 -0.000000 0.000000
O 1.236847 0.000000 -0.000000
H -0.576537 0.951580 -0.000000
H -0.576537 -0.951580 -0.000000
*
The output is similar to that from a canonical EOM-CCSD calculation:

----------------------
EOM-CCSD RESULTS (RHS)
----------------------
IROOT= 1: 0.147533 au 4.015 eV 32379.7 cm**-1

-0.418408 2 -> 8
0.418518 3 -> 8
0.228340 6 -> 8
-0.228336 7 -> 8
IROOT= 2: 0.313869 au 8.541 eV 68886.2 cm**-1
-0.399656 2 -> 9
0.399757 3 -> 9
0.259109 6 -> 9
-0.259107 7 -> 9
IROOT= 3: 0.343726 au 9.353 eV 75439.1 cm**-1
0.352815 2 -> 8
0.352629 3 -> 8
-0.266564 4 -> 8
0.266605 5 -> 8
0.175182 6 -> 8
0.175151 7 -> 8
IROOT= 4: 0.363858 au 9.901 eV 79857.7 cm**-1
-0.290406 2 -> 10
0.290481 3 -> 10
0.341790 4 -> 8
0.341754 5 -> 8
0.133101 6 -> 10
-0.133099 7 -> 10
IROOT= 5: 0.389200 au 10.591 eV 85419.5 cm**-1
-0.211907 2 -> 8
0.211950 3 -> 8
-0.414297 6 -> 8
0.414295 7 -> 8
IROOT= 6: 0.414461 au 11.278 eV 90963.7 cm**-1
-0.305034 2 -> 10
0.305113 3 -> 10
-0.267123 4 -> 8
-0.267096 5 -> 8
0.227699 6 -> 10
-0.227697 7 -> 10
IROOT= 7: 0.423449 au 11.523 eV 92936.4 cm**-1
0.411141 2 -> 11
-0.411246 3 -> 11
8.10 Multireference Configuration Interaction and Pertubation Theory 175
-0.247389 6 -> 11
0.247384 7 -> 11
IROOT= 8: 0.443945 au 12.080 eV 97434.7 cm**-1
-0.470107 4 -> 9
-0.470057 5 -> 9
IROOT= 9: 0.510256 au 13.885 eV 111988.3 cm**-1
-0.470838 4 -> 10
-0.470785 5 -> 10
The IP and EA versions can be called by using the keywords bt-PNO-IP-EOM-CCSD and bt-PNO-EA-
EOM-CCSD, respectively. Furthermore, the STEOM version can be invoked by using the keywords bt-PNO-
STEOM-CCSD
8.9.2 Capabilities
All of the features of canonical EOM-CC and STEOM-CC are available in the PNO based approaches.
8.10 Multireference Configuration Interaction and Pertubation Theory
8.10.1 Introductory Remarks
ORCA contains a multireference correlation module designed for traditional (uncontracted) approaches
(configuration interaction, MR-CI, and perturbation theory, MR-PT). For clarification, these approaches have
in common that they consider excitations from each and every configuration state function (CSF) of the
reference wavefunction. Hence, the computational cost of such approaches grows rapidly with the size of
the reference space (e.g. CAS-CI). Internally contracted on the other hand define excitations with respect
to the entire reference wavefunction and hence do not share the same bottlenecks. ORCA also features
internally contracted approaches (perturbation theory, NEVPT2 and configuration interaction, FIC-MRCI),
which are described elsewhere in the manual. The following chapter focuses on the traditional multi-reference
approaches as part of the orca mrci module.
Although there has been quite a bit of experience with it, this part of the program is still somewhat hard to
use and requires patience and careful testing before the results should be accepted. While we try to make
your life as easy as possible, you have to be aware that ultimately any meaningful multireference ab initio
calculation requires more insight and planning from the user side than standard SCF or DFT calculation
or single reference correlation approaches like MP2 so dont be fainthearted! You should also be aware
that with multireference methods it is very easy to let a large computer run for a long time and still to
not produce a meaningful result your insight is a key ingredient to a successful application! Below a few
examples illustrate some basic uses of the orca mrci module.
RI-approximation
First of all, it is important to understand that the default mode of the MR-CI module in its present
implementation performs a full integral transformation from the AO to the MO basis. This becomes very
laborious and extremely memory intensive beyond approximately 200 MOs that are included in the CI.
Alternatively, one can construct molecular electron-electron repulsion integrals from the resolution of the
identity (RI) approximation. Thus a meaningful auxiliary basis set must be provided if this option is chosen.
We recommend the fitting bases developed by the TurboMole developers for MP2 calculations. These give
accurate transition energies; however, the error in the total energies is somewhat higher and may be on the
order of 1 mEh or so. Check IntMode to change the default mode for the integral transformation. Note that
in either way, the individually selecting MRCI module requires to have all integrals in memory which sets a
limit on the size of the molecule that can be studied.
Individual Selection
Secondly, it is important to understand that the MR-CI module is of the individually selecting type. Thus,
only those excited configuration state functions (CSFs) which interact more strongly than a given threshold
(Tsel ) with the 0th order approximations to the target states will be included in the variational procedure. The
effect of the rejected CSFs is estimated using second order perturbation theory. The 0th order approximations
to the target states are obtained from the diagonalization of the reference space configurations. A further
approximation is to reduce the size of this reference space through another selection all initial references
which contribute less than a second threshold (Tpre ) to the 0th order states are rejected from the reference
space.
Single excitations
One important aspect concerns the single excitations. If the reference orbitals come from a CASSCF
calculation the matrix elements between the reference state and the single excitations vanishes and the singles
will not be selected. However, they contribute to fourth and higher orders in perturbation theory and may
be important for obtaining smooth potential energy surfaces and accurate molecular properties. Hence, the
default mode of the MRCI module requires to include all of the single excitations via the flag AllSingles
=true. This may lead to lengthy computations if the reference spaces becomes large!
Reference Spaces
Third, the reference spaces in the MR-CI module can be of the complete active space (CAS(n-electrons,m-
orbitals)) or restricted active space (RAS, explained later) type. It is important to understand that the
program uses the orbitals around the HOMO-LUMO gap as provided by the user to build up the reference
space! Thus, if the orbitals that you want to put in the active space are not coming naturally from your
SCF calculation in the right place you have to reorder them using the moread and rotate features
together with the NoIter directive. To select the most meaningful and economic reference space is the most
important step in a multireference calculation. It always requires insight from the user side and also care
and, perhaps, a little trial and error.
Size Consistency
Fourth, it is important to understand that CI type methods are not size consistent. Practically speaking
the energy of the supermolecule A-B with noninteracting A and B fragments is not equal to the energies
of isolated A and isolated B. There are approximate ways to account for this (ACPF, AQCC and CEPA
methods) but the effect will be present in the energies, the more so the more electrons are included in the
treatment. The same is not true for the perturbation theory based methods which are size consistent as long
as the reference wavefunction is.
Performance
There are many flags that control the performance of the MR-CI program. Please refer to chapter 0 for a
description of possible flags, thresholds and cut-offs. The most important thresholds are Tsel and Tpre , and
for SORCI also Tnat .
For some methods, like ACPF, it is possible to compare the performance of the MRCI module with the
performance of the MDCI module. The MDCI module has been written to provide optimum performance
if no approximations are introduced. The MRCI module has ben written more with the idea of flexibility
rather than the idea of performance. Let us compare the performance of the two programs in a slightly
nontrivial calculation the zwitter-ionic form of serine. We compare the selecting MRCI approach with the
approximation free MDCI module. The molecular size is such that still all four index integrals can be stored
in memory.
Table 8.11: Comparison of the performance of the MRCI and MDCI modules for a single reference calculation
with the bn-ANO-DZP basis set on the zwitter-ionic form of serine (14 atoms, 133 basis functions).
Module Method Tsel (Eh) Time (sec) Energy (Eh)
MRCI ACPF 106 3277 -397.943250
MDCI ACPF 0 1530 -397.946429
MDCI CCSD 0 2995 -397.934824
MDCI CCSD(T) 0 5146 -397.974239
The selecting ACPF calculation selects about 15% of the possible double excitations and solves a secular
problem of size 360,000 CSFs. The MDCI module ACPF calculation optimizes approximately 2.5 million
wavefunction amplitudes and this is not a large molecule or a large basis set! Despite the fact that the
MDCI module makes no approximation, it runs twice as fast as the selected MRCI module and an estimated
50 times faster than the unselected MRCI module! This will become even more pronounced for the larger
and more accurate basis sets that one should use in such calculations anyways. The error of the selection
is on the order of 3 mEh or 2 kcal/mol in the total energy. One can hope that at least part of this error
cancels upon taking energy differences.11 The more rigorous CCSD calculation takes about a factor of two
longer than the ACPF calculation which seems reasonable. The triples add another factor of roughly 2 in
this example but this will increase for larger calculations since it has a steeper scaling with the system size.
The ACPF energy is intermediate between CCSD and CCSD(T) which is typical ACPF overshoots the
effects of disconnected quadruples which partially compensates for the neglect of triples.
These timings will strongly depend on the system that you run the calculation on. Nevertheless, what you
should take from this example are the message that if you can use the MDCI module, do it.
The MDCI module can avoid a full integral transformation for larger systems while the MRCI module can use
selection and the RI approximation for larger systems. Both types of calculation will become very expensive
very quickly! Approximate MDCI calculations are under development.
Symmetry
The MRCI program really takes advantage of symmetry adapted orbitals. In this case the MRCI matrix
can be blocked according to irreducible representations and be diagonalized irrep by irrep. This is a big
computational advantage and allows one to converge on specific excited states much more readily than if
symmetry is not taken into account.
The syntax is relatively easy. If you specify:
newblock 1 *
nroots 8
refs cas(4,4) end
end
Then the * indicates that this is to be repeated in each irrep of the point group. Thus, in C2v the program
would calculate 8 singlet roots in each of the four irreps of the C2v point group thus leading to a total of 32
states.
Alternatively, you can calculate just a few roots in the desired irreps:
11
Depending on whether one wants to take a pessimistic or an optimistic view one could either say that this result
shows what can be achieved with a code that is dedicated to a single determinant reference. Alternatively one
could (and perhaps should) complain about the high price one pays for the generality of the MRCI approach. In
any case, the name of the game would be to develop MR approaches that are equally efficient to single reference
approaches. See FIC-MRCI chapter for more information.
newblock 1 0
nroots 3
refs cas(4,4) end
end
newblock 1 2
nroots 5
refs cas(4,4) end
end
newblock 3 1
nroots 1
refs cas(4,4) end
end
In this example, we would calculate 3 singlet roots in the irrep 0 (which is A1 ), then five roots in irrep 2
(which is B1 ) and then 1 triplet root in irrep 1 (which is B2 ).
Obviously, the results with and without symmetry will differ slightly. This is due to the fact that without
symmetry the reference space will contain references that belong to wrong symmetry but will carry with
them excited configurations of right symmetry. Hence, the calculation without use of symmetry will have
more selected CSFs and hence a slightly lower energy. This appears to be unavoidable. However, the effects
should not be very large for well designed reference spaces since the additional CSFs do not belong to the
first order interacing space.
8.10.2 A Tutorial Type Example of a MR Calculation
Perhaps, the most important use of the MR-CI module is for the calculation of transition energies and optical
spectra. Let us first calculate the first excited singlet and triplet state of the formaldehyde molecule using
the MR-CI method together with the Davidson correction to approximately account for the effect of unlinked
quadruple substitutions. We deliberately choose a somewhat small basis set for this calculation which is
already reasonable since we only look at a valence excited state and want to demonstrate the principle.
Suppose that we already know from a ground state calculation that the HOMO of H2 CO is an oxygen lone
pair orbitals and the LUMO the MO. Thus, we want to calculate the singlet and triplet n transitions
and nothing else. Consequently, we only need to correlate two electrons in two orbitals suggesting a CAS(2,2)
reference space.
# A simple MRCI example

! def2-SVP def2-SVP/C UseSym
%method frozencore fc_ewin

end
%mrci ewin -3,1000

CIType MRCI
EUnselOpt FullMP2
DavidsonOpt Davidson1
UseIVOs true
tsel 1e-6
tpre 1e-2
MaxMemInt 256
MaxMemVec 32
IntMode FullTrafo
AllSingles true
Solver Diag
# ground state 1A1
NewBlock 1 0
NRoots 1
Excitations cisd
Refs CAS(2,2) end
End
# HOMO LUMO transition 1A2
NewBlock 1 1
NRoots 1
Excitations cisd
Refs CAS(2,2) end
End
# HOMO LUMO triplet transition 3A2
NewBlock 3 1
NRoots 1
Excitations cisd
Refs CAS(2,2) end
end
end
* int 0 1
C 0 0 0 0.000000 0.000 0.000
O 1 0 0 1.200371 0.000 0.000
H 1 2 0 1.107372 121.941 0.000
H 1 2 3 1.107372 121.941 180.000
*
This input which is much more than what is really required - needs some explanations: First of all, we
choose a standard RHF calculation with the SVP basis set and we assign the SV/C fitting basis although it
is not used in the SCF procedure at all. In the %mrci block all details of the MR-CI procedure are specified.
First, EWin (%method frozencore fc ewin) selects the MOs within the given orbital energy range to be
included in the correlation treatment. The CIType variable selects the type of multireference treatment.
Numerous choices are possible and MRCI is just the one selected for this application.
NOTE: The CIType statement selects several default values for other variables. So it is
a very good idea to place this statement at the beginning of the MR-CI block and possibly
overwrite the program selected defaults later . If you place the CIType statement after one of
the values which it selects by default your input will simply be overwritten!
The variables EUnselOpt and DavidsonOpt control the corrections to the MR-CI energies. EUnselOpt specifies
the way in which the MR-CI energies are extrapolated to zero threshold TSel . Here we choose a full MR-MP2
calculation of the missing contributions to be done after the variational step, i.e. using the relaxed part of
the reference wavefunction as a 0th order state for MR-PT. The DavidsonOpt controls the type of estimate
made for the effect of higher substitutions. Again, multiple choices are possible but the most commonly
used one (despite some real shortcomings) is certainly the choice Davidson1. The flag UseIVOs instructs the
program to use improved virtual orbitals. These are virtual orbitals obtained from a diagonalization of
the Fock operator from which one electron has been removed in an averaged way from the valence orbitals.
Thus, these orbitals see only a N 1 electron potential (as required) and are not as diffuse as the standard
virtual orbitals from Hartree-Fock calculations. If you input DFT orbitals in the MR-CI moldule (which is
perfectly admittable and also recommened in some cases, for example for transition metal complexes) then
it is recommended to turn that flag off since the DFT orbitals are already o.k. in this respect. The two
thresholds Tsel and Tpre are already explained above and represent the selection criteria for the first order
interacting space and the reference space respectively. Tsel is given in units of Eh and refers to the second
order MR-MP2 energy contribution from a given excited CSF. 106 Eh is a pretty good value. Reliable
results for transition energies start with 105 ; however, the total energy is converging pretty slowly with
this parameter and this is one of the greatest drawbacks of individually selecting CI procedures! (see below).
Tpre is dimensionless and refers to the weight of a given initial reference after diagonalization of the the
given initial reference space (104 is a pretty good value and there is little need to go much lower. Aggressive
values such as 102 only select the truly leading configurations for a given target state which can be time
saving. Intermediate values are not really recommended). The parameters MaxMemInt and MaxMemVec tell
the program how much memory (in MB) it is allowed to allocate for integrals and for trial and sigma-vectors
respectively.
The flag IntMode tells the program to perform a full integral transformation. This is possible for small cases
with less than, say, 100200 MOs. In this case that it is possible it speeds up the calculations considerably.
For larger molecules you have to set this flag to RITrafo which means that integrals are recomputed on the
fly using the RI approximation which is more expensive but the only way to do the calculation. To switch
between the possible modes use:
%mrci IntMode FullTrafo # exact 4 index transformation

RITrafo # use auxiliary basis sets
For small molecules or if high accuracy in the total energies is required it is much better to use the exact four
index transformation. The limitations are that you will run out of disk space or main memory with more
than ca. 200300 MOs.
The variable Solver can be diag (for Davidson type diagonalization) or DIIS for multirrot DIIS type
treatments.
%mrci Solver Diag # Davidson solver

DIIS # Multiroot DIIS
For CI methods, the diag solver is usually preferable. For methods like ACPF that contain nonlinear terms,
DIIS is imperative.
Next in the input comes the definition of what CI matrices are to be constructed and diagonalized. Each
multiplicity defines a block of the CI matrix which is separately specified. Here we ask for two blocks singlet
and triplet. The general syntax is:
NewBlock Multiplicity Irrep

NRoots 1 # Number of roots to determine
Excitations cisd # Type of excitations
Refs CAS(NEl,NOrb) end # Reference space def.
end # Finalize the block
Now that all input is understood let us look at the outcome of this calculation:
The first thing that happens after the SCF calculation is the preparation of the frozen core Fock matrix and
the improved virtual orbitals by the program orca ciprep. From the output the energies of the IVOs can be
seen. In this case the LUMO comes down to 8.2 eV which is much more reasonable than the SCF value of
+3. . . . eV. Concomitantly, the shape of this MO will be much more realistic and this important since this
orbital is in the reference space!
------------------------------------------------------------------------------
ORCA CI-PREPARATION
------------------------------------------------------------------------------
One-Electron Matrix ... Test-SYM-MRCI-H2CO.H.tmp

GBW-File ... Test-SYM-MRCI-H2CO.gbw
Improved virtual orbitals ... Test-SYM-MRCI-H2CO.ivo
First MO in the CI ... 2
Internal Fock matrix ... Test-SYM-MRCI-H2CO.cif.tmp
LastInternal Orbital ... 6
Integral package used ... LIBINT
Reading the GBW file ... done
Symmetry usage ... ON
Reading the one-electron matrix ... done

Forming inactive density ... done
Forming averaged valence density ...
Scaling the occupied orbital occupation numbers
First MO ... 2
Last MO ... 7
Number of electrons in the range ... 12
Scaling factor ... 0.917
done
Forming internal density ... done
Forming Fock matrix/matrices ...
Nuclear repulsion ... 31.371502
Core repulsion ... 31.371502
One-electron energy ... -114.942082
Fock-energy ... -94.993430
Final value ... -73.596255
done
Modifying virtual orbitals ...
Last occupied MO ... 7
Total number of MOs ... 38
Number of virtual MOs ... 30
Doing diagonalization with symmetry
The improved virtual eigenvalues:
0: -0.2955 au -8.041 eV 2- B1
1: -0.0701 au -1.908 eV 6- A1
2: -0.0176 au -0.480 eV 3- B2
3: 0.0064 au 0.175 eV 7- A1
4: 0.2922 au 7.951 eV 8- A1
5: 0.2948 au 8.021 eV 3- B1
6: 0.3836 au 10.439 eV 4- B2
7: 0.4333 au 11.790 eV 9- A1
8: 0.4824 au 13.128 eV 5- B2
9: 0.5027 au 13.680 eV 10- A1
10: 0.7219 au 19.643 eV 11- A1
11: 0.8351 au 22.724 eV 4- B1
12: 0.9372 au 25.501 eV 6- B2
13: 1.0265 au 27.932 eV 1- A2
14: 1.1141 au 30.317 eV 12- A1
15: 1.2869 au 35.017 eV 5- B1
16: 1.4605 au 39.743 eV 7- B2
...
done
Transforming integrals ... done
Storing passive energy ... done ( -73.59625452 Eh)
Transforming internal FI ... done
.... done with the Frozen Core Fock matrices
The next step is to transform the electron-electron repulsion integrals into the MO basis:
------------------------------
PARTIAL COULOMB TRANSFORMATION
------------------------------
Dimension of the basis ... 38

Number of internal MOs ... 36 (2-37)
Pair cutoff ... 1.000e-11 Eh
Number of AO pairs included in the trafo ... 741
Total Number of distinct AO pairs ... 741
Memory devoted for trafo ... 256.0 MB
Max. Number of MO pairs treated together ... 45282
Max. Number of MOs treated per batch ... 36
Number Format for Storage ... Double (8 Byte)
Integral package used ... LIBINT
--->>> The Coulomb operators (i,j|mue,nue) will be calculated
Starting integral evaluation:

<ss|**>: 9404 b 1 skpd 0.023 s ( 0.002 ms/b)
<sp|**>: 10260 b 0 skpd 0.030 s ( 0.003 ms/b)
<sd|**>: 3420 b 0 skpd 0.016 s ( 0.005 ms/b)
<pp|**>: 3591 b 0 skpd 0.026 s ( 0.007 ms/b)

<pd|**>: 2052 b 0 skpd 0.025 s ( 0.012 ms/b)
<dd|**>: 513 b 0 skpd 0.009 s ( 0.017 ms/b)
Collecting buffer AOJ
... done with AO integral generation
Closing buffer AOJ ( 0.00 GB; CompressionRatio= 4.22)
Number of MO pairs included in the trafo ... 666
... Now sorting integrals
IBATCH = 1 of 2
IBATCH = 2 of 2
Closing buffer JAO ( 0.00 GB; CompressionRatio= 5.20)
TOTAL TIME for half transformation ... 0.324 sec
AO-integral generation ... 0.118 sec
Half transformation ... 0.059 sec
J-integral sorting ... 0.146 sec
Collecting buffer JAO
-------------------
FULL TRANSFORMATION
-------------------
Processing MO 10
Processing MO 20
Processing MO 30
full transformation done
Number of integrals made ... 222111
Number of integrals stored ... 59070
Timings:
Time for first half transformation ... 0.326 sec
Time for second half transformation ... 0.160 sec
Total time ... 0.516 sec
This will result in a few additional disk files required by orca mrci. The program then tells you which
multiplicities will be treated in this MRCI run:
------------------
CI-BLOCK STRUCTURE
------------------
Number of CI-blocks ... 3
===========
CI BLOCK 1
===========
Multiplicity ... 1
Irrep ... 0
Number of reference defs ... 1
Reference 1: CAS(2,2)
Excitation type ... CISD

Excitation flags for singles:
1 1 1 1
Excitation flags for doubles:
1 1 1 / 1 1 1 / 1 1 1
===========
CI BLOCK 2
===========
Multiplicity ... 1
Irrep ... 1

1 1 1 1
1 1 1 / 1 1 1 / 1 1 1
===========
CI BLOCK 3
===========
Multiplicity ... 3
Irrep ... 1

1 1 1 1
1 1 1 / 1 1 1 / 1 1 1
--------------------------------------------------------------------
-------------------- ALL SETUP TASKS ACCOMPLISHED ------------------
-------------------- ( 1.512 sec) ------------------
--------------------------------------------------------------------
Now that all the setup tasks have been accomplished the MRCI calculation itself begins.
###################################################
# #
# M R C I #
# #
# TSel = 1.000e-06 Eh #
# TPre = 1.000e-02 #
# TIntCut = 1.000e-10 Eh #
# Extrapolation to unselected MR-CI by full MP2 #
# DAVIDSON-1 Correction to full CI #
# #
###################################################
---------------------
INTEGRAL ORGANIZATION
---------------------
Reading the one-Electron matrix ... done

Reading the internal Fock matrix ... assuming it to be equal to the one-electron matrix!!!
done
Preparing the integral list ... done

Loading the full integral list ... done
Making the simple integrals ... done
***************************************
* CI-BLOCK 1 *
***************************************
Configurations with insufficient # of SOMOs WILL be rejected

Building a CAS(2,2) for multiplicity 1 and irrep=A1
Reference Space:
Initial Number of Configurations : 2
Internal Orbitals : 2 - 6
Active Orbitals : 7 - 8
External Orbitals : 9 - 37
The number of CSFs in the reference is 2
Calling MRPT_Selection with N(ref)=2
In the first step, the reference space is diagonalized. From this CI, the most important configurations are
selected with Tpre:
------------------
REFERENCE SPACE CI
------------------
Pre-diagonalization threshold : 1.000e-02

N(ref-CFG)=2 N(ref-CSF)=2
****Iteration 0****
Lowest Energy : -113.779221544551
Maximum Energy change : 113.779221544551 (vector 0)
Maximum residual norm : 0.000000000000
*** CONVERGENCE OF RESIDUAL NORM REACHED ***

Reference space selection using TPre= 1.00e-02
... found 1 reference configurations (1 CSFs)

... now redoing the reference space CI ...
N(ref-CFG)=1 N(ref-CSF)=1
****Iteration 0****
Lowest Energy : -113.778810020485
In this case, the CAS space only has 2 correctly symmetry adapted CSFs one of which (the closed-shell
determinant) is selected. In general, larger CAS spaces usually carry around a lot of unnecessary CSFs which
are not needed for anything and then the selection is important to reduce the computational effort. The
result of the second reference space CI is printed:
----------
CI-RESULTS
----------
The threshold for printing is 0.3 percent

The weights of configurations will be printed. The weights are summed over
all CSFs that belong to a given configuration before printing

1.0000 : h---h---[20]
Energy is the total energy in Eh. In the present case we can compare to the SCF energy -113.778810021
Eh and find that the reference space CI energy is identical, as it has to be since the lowest state coincides
with the reference space. RefWeight gives the weight of the reference configurations in a CI state. This is
1.0 in the present case since there were only reference configurations. The number 1.000 is the weight of
the following configuration in the CI vector. The description of the configuration h---h---[20]p---p--- is
understood as follows:12 The occupation of the active orbitals is explicitly given in square brackets. Since the
HOMO orbitals is number 7 from the SCF procedure, this refers to MOs 7 and 8 in the present example since
we have two active orbitals. The 2 means doubly occupied, the 0 means empty. Any number (instead of ---)
appearing after an h gives the index of an internal orbital in which a hole is located. Simarly, any number
after a p gives the index of an virtual (external) MO where a particle is located. Thus h---h---[20] is a
closed shell configuration and it coincides with the SCF configurationthis was of course to be expected. The
second root (in CI-Block 2) h---h---[11] by comparison refers to the configuration in which one electron
has been promoted from the HOMO to the LUMO and is therefore the desired state that we wanted to
calculate. Things are happy therefore and we can proceed to look at the output.
The next step is the generation of excited configurations and their selection based on Tsel:
------------------------------
MR-PT SELECTION TSel= 1.00e-06
------------------------------
Setting reference configurations WITH use of symmetry

Building active patterns WITH use of symmetry
Selection will be done from 1 spatial configurations

Selection will make use of spatial symmetry
( 0) Refs : Sel: 1CFGs/ 1CSFs Gen: 1CFGs/ 1CSFs
Building active space densities ... done
Building active space Fock operators ... done
( 1) (p,q)->(r,s): Sel: 1CFGs/ 1CSFs Gen: 1CFGs/ 1CSFs
( 2) (i,-)->(p,-): Sel: 1CFGs/ 1CSFs Gen: 1CFGs/ 1CSFs
( 3) (i,j)->(p,q): Sel: 8CFGs/ 8CSFs Gen: 8CFGs/ 8CSFs
( 4) (i,p)->(q,r): Sel: 0CFGs/ 0CSFs Gen: 1CFGs/ 1CSFs
( 5) (p,-)->(a,-): Sel: 8CFGs/ 8CSFs Gen: 8CFGs/ 8CSFs
( 6) (i,-)->(a,-): Sel: 52CFGs/ 52CSFs Gen: 52CFGs/ 52CSFs
( 7) (i,j)->(p,a): Sel: 95CFGs/ 166CSFs Gen: 96CFGs/ 167CSFs
( 8) (i,p)->(q,a): Sel: 21CFGs/ 42CSFs Gen: 22CFGs/ 44CSFs
12
Note that for printing we always sum over all linearly independent spin couplings of a given spatial configuration and
only print the summed up weight for the configuration rather than for each individual CSF of the configuration.
( 9) (p,q)->(r,a): Sel: 3CFGs/ 3CSFs Gen: 5CFGs/ 5CSFs

(10) (i,p)->(a,b): Sel: 555CFGs/ 1082CSFs Gen: 584CFGs/ 1139CSFs
(11) (p,q)->(a,b): Sel: 124CFGs/ 124CSFs Gen: 148CFGs/ 148CSFs
(12) (i,j)->(a,b): Sel: 1688CFGs/ 2685CSFs Gen: 1887CFGs/ 2947CSFs
Selection results:
Total number of generated configurations: 2814
Number of selected configurations : 2557 ( 90.9%)
Total number of generated CSFs : 4522
Number of selected CSFS : 4173 ( 92.3%)
The selected tree structure:

Number of selected Internal Portions : 11
Number of selected Singly External Portions: 27
average number of VMOs/Portion : 6.39
percentage of selected singly externals : 22.83
Number of selected Doubly External Portions: 21
average number of VMOs/Portion : 107.59
percentage of selected doubly externals : 27.76
Here, the program loops through classes of excitations. For each excitation it produces the excited con-
figurations (CFGs) and from it the linearly independent spin functions (CSFs) which are possible within
the configuration. It then calculates the interaction with the contracted 0th order roots and includes all
CSFs belonging to a given CFG in the variational space if the largest second order perturbation energy is
larger or equal to Tsel. In the present case 136,000 CSFs are produced of which 25% are selected. For
larger molecules and basis sets it is not uncommon to produce 109 1010 configurations and then there is no
choice but to select a much smaller fraction than 20%. For your enjoyment, the program also prints the total
energies of each state after selection:
Diagonal second order perturbation results:

State E(tot) E(0)+E(1) E2(sel) E2(unsel)
Eh Eh Eh Eh
----------------------------------------------------------------
0 -114.108347273 -113.778810020 -0.329430 -0.000107
You can ignore this output if you want. In cases that the perturbation procedure is divergent (not that
uncommon!) the total energies look strangedont worrythe following variational calculation is still OK.
The second order perturbation energy is here divided into a selected part E2(sel) and the part procedure by
the unselected configurations E2(unsel). Depending on the mode of EUnselOpt this value may already be
used later as an estimate of the energetic contribution of the unselected CSFs.13
Now we have 4,200 CSFs in the variational space of CI block 1 and proceed to diagonalize the Hamiltonian
over these CSFs using a Davidson or DIIS type procedure:
13
In this case the maximum overlap of the 0th order states with the final CI vectors is computed and the perturbation
energy is added to the most similar root. This is of course a rather crude approximation and a better choice is
to recomputed the second order energy of the unselected configurations rigorously as is done with EUnselOpt =
FullMP2.
------------------------
DAVIDSON-DIAGONALIZATION
------------------------
Dimension of the eigenvalue problem ... 4173

Number of roots to be determined ... 1
Maximum size of the expansion space ... 15
Convergence tolerance for the residual ... 1.000e-06
Convergence tolerance for the energies ... 1.000e-06
Orthogonality tolerance ... 1.000e-14
Level Shift ... 0.000e+00
Constructing the preconditioner ... o.k.
Building the initial guess ... o.k.
Number of trial vectors determined ... 2
****Iteration 0****
Size of expansion space: 2
Lowest Energy : -113.854262408162
****Iteration 1****
Lowest Energy : -114.076119460817
****Iteration 2****
Lowest Energy : -114.085249547769
****Iteration 3****
Lowest Energy : -114.086014164840
****Iteration 4****
Lowest Energy : -114.086071121272
****Iteration 5****
Lowest Energy : -114.086076153851
****Iteration 6****
Lowest Energy : -114.086076506777
*** CONVERGENCE OF ENERGIES REACHED ***
Storing the converged CI vectors ... Test-SYM-MRCI-H2O.mrci.vec
*** DAVIDSON DONE ***

Returned from DIAG section
The procedure converges on all roots simultaneously and finishes after six iterations which is reasonable. Now
the program calculates the Davidson correction (DavidsonOpt) which is printed for each root.
Davidson type correction:

Root= 0 W= 0.912 E0= -113.778810020 ECI= -114.086076507 DE=-0.026914
Already in this small example the correction is pretty large, ca. 27 mEh for the ground state (and 36
mEh for the excited state, later in the output). Thus, a contribution of 9 mEh = 0.25 eV is obtained for
the transition energy which is certainly significant. Unfortunately, the correction becomes unreliable as the
reference space weight drops or the number of correlated electrons becomes large. Here 0.912 and 0.888 are
still OK and the system is small enough to expect good results from the Davidson correction.
The next step is to estimate the correction for the unselected configurations:
Unselected CSF estimate:

Full relaxed MR-MP2 calculation ...
Selection will be done from 1 spatial configurations

done
Selected MR-MP2 energies ...
Root= 0 E(unsel)= -0.000106951
In the present case this is below 1 mEh and also very similar for all three states such that it is not important
for the transition energy.
----------
CI-RESULTS
----------


0.9124 : h---h---[20]
0.0114 : h 6h 6[22]
The final ground state energy is -114.113097002 which is an estimate of the full CI energy in this basis set.
The leading configuration is still the closed-shell configuration with a weight of 91%. However, a double
excitation outside the reference space contributes some 1%. This is the excitation MO6,MO6 LUMO,LUMO.
This indicates that more accurate results are expected once MO6 is also included in the reference space (this
is the HOMO-1). The excited state is dominated by the HOMO-LUMO transition (as desired) but a few
other single- and double- excitations also show up in the final CI vector.
Now that all CI vectors are known we can order the states according to increasing energy and print (vertical)
transition energies:
-------------------
TRANSITION ENERGIES
-------------------
The lowest energy is -114.113097002 Eh

0 1 A1 0 0 0.000 0.000 0.0
1 3 A2 0 2 134.073 3.648 29425.7
2 1 A2 0 1 148.490 4.041 32589.8
This result is already pretty good and the transition energies are within 0.1 eV of their experimental gas
phase values ( 3.50 and 4.00 eV) and may be compared to the CIS values of 3.8 and 4.6 eV which are
considerably in error.
In the next step the densities and transition densities are evaluated and the absorption and CD spectra
are calculated (in the dipole length formalism) for the spin-allowed transitions together with state dipole
moments:
------------------------------------------------------------------------------------------
ABSORPTION SPECTRUM
------------------------------------------------------------------------------------------
States Energy Wavelength fosc T2 TX TY TZ
(cm-1) (nm) (D**2) (D) (D) (D)
------------------------------------------------------------------------------------------
0( 0)-> 0( 1) 1 32589.8 306.8 0.000000000 0.00000 -0.00000 0.00000 0.00000
------------------------------------------------------------------------------
CD SPECTRUM
------------------------------------------------------------------------------
States Energy Wavelength R*T RX RY RZ
(cm-1) (nm) (1e40*sgs) (au) (au) (au)
------------------------------------------------------------------------------
0( 0)-> 0( 1) 1 32589.8 306.8 0.00000 -0.00000 -0.00000 1.18711
------------------------------------------------------------------------------
STATE DIPOLE MOMENTS
------------------------------------------------------------------------------
Root Block TX TY TZ |T|
(Debye) (Debye) (Debye) (Debye)
------------------------------------------------------------------------------
0 0 0.00000 -0.00000 2.33244 2.33244
0 2 0.00000 -0.00000 1.45831 1.45831
0 1 0.00000 -0.00000 1.58658 1.58658
Here the transition is symmetry forbidden and therefore has no oscillator strength. The state dipole moment
for the ground state is 2.33 Debye which is somewhat lower than 2.87 Debye from the SCF calculation. Thus,
the effect of correlation is to reduce the polarity consistent with the interpretation that the ionicity of the
bonds, which is always overestimated by HF theory, is reduced by the correlation. Finally, you also get a
detailed population analysis for each generated state density which may be compared to the corresponding
SCF analysis in the preceding part of the output.
This concludes the initial example on the use of the MR-CI module. The module leaves several files on disk
most of which are not yet needed but in the future will allow more analysis and restart and the like. The
.ivo file is a standard .gbw type file and the orbitals therein can be used for visualization. This is important
in order to figure out the identity of the generated IVOs. Perhaps they are not the ones you wanted and then
you need to re-run the MR-CI with the IVOs as input, NoIter and the IVO feature in the new run turned
off! We could use the IVOs as input for a state averaged CASSCF calculation:
! moread UseSym KDIIS

%moinp "Test-SYM-MRCI-H2CO.ivo"
%casscf nel 2
norb 2
irrep 0,1,1
mult 1,1,3
nroots 1,1,1
end
If we based a MR-ACPF calculation on this reference space we will find that the calculated transition energies
are slightly poorer than in the MRCI+Q calculation. This is typical of approximate cluster methods that
usually require somewhat larger reference spaces for accurate results. A similar result is obtained with
SORCI.
%mrci CIType SORCI

tsel 1e-6
tpre 1e-4
tnat 1e-5
AllSingles true
doNatOrbs true
IntMode FullTrafo
# ground state 1A1
NewBlock 1 0
NRoots 1
Excitations cisd
Refs CAS(2,2) end
End
NewBlock 1 1
NRoots 1
Excitations cisd
Refs CAS(2,2) end
End
NewBlock 3 1
NRoots 1
Excitations cisd
Refs CAS(2,2) end
end
This gives:

0 1 A1 0 0 0.000 0.000 0.0
1 3 A2 0 2 146.507 3.987 32154.5
2 1 A2 0 1 161.801 4.403 35511.3
This is systematically 0.4 eV too high. But let us look at the approximate average natural orbital (AANOs)
occupation numbers:
------------------------
AVERAGE NATURAL ORBITALS
------------------------
Trace of the density to be diagonalized = 12.000000

Sum of eigenvalues = 12.000000
Natural Orbital Occupation Numbers:
N[ 2] ( A1)= 1.99832583
N[ 3] ( A1)= 1.99760289
N[ 4] ( A1)= 1.99481021
N[ 5] ( B2)= 1.99044471
N[ 6] ( B1)= 1.95799339
N[ 7] ( B2)= 1.33003795
N[ 8] ( B1)= 0.70704982
N[ 9] ( B2)= 0.00988857
N[ 10] ( A1)= 0.00448885
This shows that there is a low-occupancy orbital (MO6) that has not been part of the reference space. Thus,
we try the same calculation again but now with one more active orbital and two more active electrons:
! moread
%moinp "Test-SYM-MRCI-H2CO.gbw"
%casscf nel 4
norb 3
irrep 0,1,1
mult 1,1,3
nroots 1,1,1
end
%mrci CIType SORCI

tsel 1e-6
tpre 1e-4
tnat 1e-5
AllSingles true
doNatOrbs true
IntMode FullTrafo
# ground state 1A1
NewBlock 1 0
NRoots 1
Excitations cisd
Refs CAS(4,3) end
End
NewBlock 1 1
NRoots 1
Excitations cisd
Refs CAS(4,3) end
End
NewBlock 3 1
NRoots 1
Excitations cisd
Refs CAS(4,3) end
end
This gives:

0 1 A1 0 0 0.000 0.000 0.0
1 3 A2 0 2 137.652 3.746 30211.1
2 1 A2 0 1 153.128 4.167 33607.7
Which is now fine since all essential physics has been in the reference space. Inspection of the occupation
numbers show that there is no suspicious orbital any more. Note that this is still a much more compact
calculation that the MRCI+Q.
Likewise, we get an accurate result from MRACPF with the extended reference space.

0 1 A1 0 0 0.000 0.000 0.0
1 3 A2 0 2 134.985 3.673 29625.8
2 1 A2 0 1 148.330 4.036 32554.6
However, the SORCI calculation is much more compact. For larger molecules the difference becomes more and
more pronounced and SORCI or even MRDDCI2 (with or without +Q) maybe the only feasible methodsif
at all.
8.10.3 Excitation Energies between Different Multiplicities
As an example for a relatively accurate MRCI+Q calculation consider the following job which calculates the
triplet- ground and as the first excited singlet states of O2 .
! ano-pVQZ RI-AO cc-pVQZ/JK VeryTightSCF NoPop Conv UseSym RI-MP2 PModel

%mp2 density relaxed natorbs true end
%base "O2"
* xyz 0 3
O 0 0 0
O 0 0 1.2
*
$new_job
! ano-pVQZ RI-AO cc-pVQZ/JK VeryTightSCF NoPop Conv UseSym KDIIS
! moread
%moinp "O2.mp2nat"
%casscf nel 8
norb 6
irrep 1,0,1
nroots 1,2,1
mult 3,1,1
trafostep ri
switchstep nr
end
%mrci citype mrci

tsel 1e-7
tpre 1e-5
newblock 3 1 nroots 1 refs cas(8,6) end end
end
* xyz 0 3
O 0 0 0
O 0 0 1.2
*
Note that the linear molecule is run in D2h . This creates a slight problem as the CASSCF procedure
necessarily breaks the symmetry of the 1 state.
LOWEST ROOT (ROOT 0, MULT 3, IRREP B1g) = -149.765383866 Eh -4075.323 eV
STATE ROOT MULT IRREP DE/a.u. DE/eV DE/cm**-1

1: 0 1 B1g 0.033334 0.907 7316.0
2: 0 1 Ag 0.033650 0.916 7385.3
3: 1 1 Ag 0.062381 1.697 13691.1
The result of the MRCI+Q is:
-------------------
TRANSITION ENERGIES
-------------------

0 3 B1g 0 0 0.000 0.000 0.0
1 1 B1g 0 2 36.971 1.006 8114.2
2 1 Ag 0 1 38.021 1.035 8344.7
3 1 Ag 1 1 62.765 1.708 13775.2
These excitation energies are accurate to within a few hundred wavenumbers. Note that the 200 wavenumber
splitting in the degenerate 1 state is due to the symmetry breaking of the CAS and the individual selection.
Repeating the calculation with the MP2 natural orbitals gives an almost indistinguishable result and a ground
state energy that is even lower than what was found with the CASSCF orbitals. Thus, such natural orbitals
(that might often be easier to get) are a good substitute for CASSCF orbitals and at the same time the
symmetry breaking due to the use of symmetry appears to be difficult to avoid.
-------------------
TRANSITION ENERGIES
-------------------

0 3 B1g 0 0 0.000 0.000 0.0
1 1 B1g 0 2 37.369 1.017 8201.5
2 1 Ag 0 1 38.237 1.040 8392.1
3 1 Ag 1 1 62.731 1.707 13767.9
8.10.4 Correlation Energies
The logic we are following here is the following: CID minus SCF gives the effect of the doubles; going to CISD
gives the effect of the singles; QCISD(=CCD) minus CID gives the effect of the disconnected quadruples.
QCISD minus QCID gives simultaneously the effect of the singles and the disconnected triples. They are a bit
difficult to separate but if one looks at the singles alone and compares with singles + disconnected triples, a
fair estimate is probably obtained. Finally, QCISD(T) minus QCISD gives the effect of the connected triples.
One could of course also use CCSD instead of QCISD but I felt that the higher powers of T1 obscure the
picture a little bitbut this is open to discussion of course.
First H2 O/TZVPP at its MP2/TZVPP equilibrium geometry (Tpre =106 and Tsel =109 Eh for the MRCI
and MRACPF calculations):
Excitation class Energy (Eh) Delta-Energy (mEh)

None (RHF) -76.0624
Doubles (CID) -76.3174 255
+Singles (CISD) -76.3186 1
+Disconnected Quadruples (QCID) -76.3282 11
+Disconnected Triples (QCISD) -76.3298 2
+Connected Triples (QCISD(T)) -76.3372 7
CASSCF(8,6) -76.1160
CASSCF(8,6) + MRCI -76.3264 210
CASSCF(8,6) + MRCI+Q -76.3359 10
CASSCF(8,6) + MRACPF -76.3341 218
One observes quite good agreement between single- and multireference approaches. In particular, the
contribution of the disconnected triples and singles is very small. The estimate for the disconnected
quadruples is fairly good from either the multireference Davidson correction or the ACPF and the agreement
between CCSD(T) and these MR methods is 2-3 mEh in the total energy which is roughly within chemical
accuracy.
In order to also have an open-shell molecule let us look at NH with a N-H distance of 1.0
A using the TZVPP
basis set.
Excitation class Energy (Eh) Delta-Energy (mEh)

None (UHF) -54.9835
Doubles (CID) -55.1333 150
+Singles (CISD) -55.1344 1
+Disconnected Quadruples (QCID) -55.1366 3
+Disconnected Triples (QCISD) -55.1378 1
+Connected Triples (QCISD(T)) -55.1414 4
CASSCF(6,5) -55.0004
CASSCF(6,5) + MRCI -55.1373 137
CASSCF(6,5) + MRCI+Q -55.1429 6
CASSCF(6,5) + MRACPF -55.1413 141
Again, the agreement is fairly good and show that both single- and multiple reference approaches converge to
the same limit.
8.10.5 Thresholds
Now we choose the CO molecule (1.128 A ngstrom) with the SVP basis set and study the convergence of
the results with respect to the selection threshold. Comparison to high level single-reference approaches is
feasible (The SCF energy is -112.645 946 Eh).
8.10.5.1 Reference Values for Total Energies
The single-reference values are:
BD: -112.938 48002

CCSD: -112.939 79145
QCISD: -112.941 95700
BD(T): -112.950 17278
CCSD(T): -112.950 63889
QCISD(T): -112.951 37425
MP4(SDTQ): -112.954 80113
The calculations without connected triples (BD, CCSD, QCISD) are about the best what can be achieved
without explicitly considering triple excitations. The CCSD is probably the best in this class. As soon as
connected triples are included the CCSD(T), QCISD(T) and BD(T) values are close and from experience they
are also close to the full CI values which is then expected somewhere between 112.950 and 112.952 Eh.
8.10.5.2 Convergence of Single Reference Approaches with Respect to Tsel
Next it is studied how these single reference methods converge with Tsel :
Closed-Shell ACPF:
Tsel Energy (NCSF) Energy (NCSF)
(Eh) AllSingles=true AllSingles=false
TSel=0 -112.943 387 (5671)
TSel=1e-14 -112.943 387 (2543) -112.943 387 (2478)
TSel=1e-10 -112.943 387 (2543) -112.941 023 (2453)
TSel=1e-08 -112.943 387 (2451) -112.937 087 (2346)
TSel=1e-06 -112.943 350 (2283) -112.937 046 (2178)
TSel=1e-05 -112.943 176 (1660) -112.936 821 (1555)
TSel=1e-04 -112.944 039 ( 782) -112.938 381 ( 677)
It is clear that the convergence is erratic if the singles are not automatically included. This is the reason
for making this the default from release 2.6.35 on. In the present case singles will only be selected due
to round-off errors since by Brillouins theorem the singles have zero-interaction with the ground state
determinant. Thus, for individually selecting single-reference methods it is a good idea to automatically
include all single-excitations in order to get converged results. The alternative would be a different singles
selection procedure which has not yet been developed however. The selection of doubles appear to converge
the total energies reasonably well. It is seen that the selection selects most CSFs between 105 and 107 Eh.
Already a threshold of 106 Eh yields an error of less than 0.1 mEh which is negligible in relation to reaction
energies and the like. Even 105 Eh gives an error of less than 0.1 kcal/mol.
8.10.5.3 Convergence of Multireference Approaches with Respect to Tpre
We next turn to multireference treatments. Here we want to correlate all valence electrons in all valence
orbitals and therefore a CAS(10,8) is the appropriate choice. We first ask for the converged value of Tpre by
using Tsel =1014 and obtain for MRCI+Q:
TPre = 1e-1: -112.943 964

1e-2: -112.952 963
1e-3: -112.953 786
1e-4: -112.954 019
1e-5: -112.954 336
1e-6: -112.954 416
1e-7: -112.954 440
Thus, pretty good convergence is obtained for Tpre = 104 106 . Hence 104 is the default.
To show a convenient input consider the following:
#
# Here we calculate the CO ground state correlation energy with several methods
#
! RHF aug-SVP aug-SV/C RI-MP2 CCSD(T)
%base "1"
%mp2 density relaxed

donatorbs true
end
* int 0 1
C 0 0 0 0.000000 0.000 0.000
O 1 0 0 1.128 0.000 0.000
*
$new_job
! RHF aug-SVP MRACPF

! moread
%moinp "1.mp2nat"
# the CASSCF is done with MP2 natural orbitals which is a good idea and
# secondly we use a large level shift in order to help convergence
%casscf nel 10
norb 8
mult 1
nroots 1
shiftup 2
shiftdn 2
end
%mrci tsel 1e-8

tpre 1e-6
end
* int 0 1
C 0 0 0 0.000 0.000 0.000
O 1 0 0 1.128 0.000 0.000
*
This job computes at the same time all of the below and demonstrates once more the agreement between
consequent single- and multireference correlation methods
SCF = -112.6523
RI-MP2 = -112.9591
CCSD = -112.9604
CCSD(T) = -112.9730
CASSCF(10,8) = -112.7829
MRACPF = -112.9722
8.10.6 Energy Differences - Bond Breaking
For the calculation of energy differences we start again with the reference CCSD(T) calculation; this method
is one of the few which can claim chemical accuracy in practical applications:
Reference Total Energies for N2 at 1.0977 Angstr

om with
The SVP basis
E(CCSD) = -109.163 497
E(CCSD(T))= -109.175 625
Nitrogen Atom (4S), SVP basis, unrestricted
E(CCSD) = -54.421 004
E(CCSD(T))= -54.421 7183
Energy Difference:
Delta-E(CCSD) = -0.321 489 = 8.75 eV
Delta-E(CCSD(T))= -0.332 188 = 9.04 eV
The basis set is of course not suitable for quantitative comparison to experimental values. However, this is
not the point here in these calculations which are illustrative in nature. The SVP basis is just good enough
to allow for a method assessment without leading to excessively expensive calculations.
This is now to be compared with the corresponding energy differences computed with some single-reference
approaches. A typical input is (this is a somewhat old-fashioned example in the present program version
you would do a full valence CASSCF(10,8) or CASSCF(6,6) and invoke the MR-methods with a single
keyword):
! RHF SVP TZVPP/C VeryTightSCF NoPop
%base "1"
* xyz 0 1
N 0 0 0
N 0 0 1.0977
*
end
%mrci EWin -3,1000

CIType MRACPF2a
Solver DIIS
IntMode FullTrafo
UseIVOs true
AllSingles true
TSel 1e-14
TPre 1e-05
TNat 0.0
ETol 1e-10
RTol 1e-10
NewBlock 1 *
NRoots 1
Excitations CISD
refs CAS(0,0) end
end
end
$new_job
! ROHF SVP TZVPP/C VeryTightSCF NoPop PModel
%base "2"
* xyz 0 4
N 0 0 0
*
end
%mrci EWin -3,1000

CIType MRACPF2a
IntMode FullTrafo
UseIVOs true
AllSingles true
TSel 1e-14
TPre 1e-05
TNat 0.0
ETol 1e-10
RTol 1e-10
NewBlock 4 *
NRoots 1
Excitations CISD
refs CAS(3,3) end
end
end
The results are:
Single reference approaches:

Method N2-Molecule N-Atom Delta-E
CISD+Q : -109.167 904 -54.422 769 8.77 eV
ACPF : -109.166 926 -54.421 783 8.80 eV
ACPF2 : -109.166 751 -54.421 333 8.82 eV
ACPF2a : -109.166 730 -54.421 186 8.83 eV
CEPA1 : -109.159 721 -54.422 564 8.56 eV
CEPA2 : -109.172 888 -54.422 732 8.91 eV
CEPA3 : -109.161 034 -54.422 589 8.59 eV
AQCC : -109.160 574 -54.420 948 8.67 eV
CEPA-0 : -109.174 924 -54.422 951 8.95 eV
With exception is CEPA1 and CEPA3, the results are OK. The reason for the poor performance of these
methods is simply that the formalism implemented is only correct for closed shells open shells require a
different formalism which we do not have available in the MRCI module (but in the single reference MDCI
module). Due to the simple approximations made in CEPA2 it should also be valid for open shells and the
numerical results are in support of that.
Next we turn to the multireference methods and take a CAS(10,8) reference as for CO in order to correlate
all valence electrons. 14
14
Most of these results have been obtained with a slightly earlier version for which the MR energies are a little
different from that what the present version gives. The energy differences will not be affected.
Multi reference approaches:

Method N2-Molecule N-Atom Delta-E
MRCISD+Q: -109.180 089 -54.422 667 9.11 eV
MRACPF : -109.178 708 -54.421 685 9.12 eV
MRACPF2 : -109.177 140 -54.421 236 9.11 eV
MRAQCC : -109.175 947 -54.420 851 9.10 eV
SORCI : -109.179 101 -54.422 703 9.08 eV
This test calculation pleasingly shows the high consistency of multireference approaches which all converge
more or less to the same result which must be accurate.
8.10.7 Energy Differences - Spin Flipping
There are a number if interesting situations in which one is interested in a small energy difference which
arises from two states of different multiplicity but same orbital configuration. This is the phenomenon met in
diradicals or in magnetic coupling in transition metal complexes. As a primitive model for such cases one
may consider the hypothetical molecule H-Ne-H in a linear configuration which will be used as a model in
this section.
The reference value is obtained by a MR-ACPF calculation with all valence electrons active (again, this
example is somewhat old fashioned in the present program version you would do a CASSCF calculation
followed by MR methods with a single keyword):
! ROHF SVP TZVPP/C VeryTightSCF NoPop

%basis aux auto
newauxgto h "TZVPP/C" end
end
* xyz 0 3
H 0 0 0
Ne 0 0 2.0
H 0 0 4.0
*
end
%mrci EWin -3,1000

CIType MRACPF2a
IntMode FullTrafo
Solver DIIS
UseIVOs true
TSel 0
TPre 1e-10
ETol 1e-09
RTol 1e-09
DoDDCIMP2 true
NewBlock 1 *
NRoots 1
Excitations CISD
refs CAS(10,6) end
end
NewBlock 3 *
NRoots 1
Excitations CISD
refs CAS(10,6) end
end
end
which gives the reference value 108 cm1 . We now compare that to several other methods which only have
the two magnetic orbitals (the 1ss on the hydrogens) in the active space:
... same as above

%mrci EWin -10,1000
CIType MRDDCI3
... same as previously
NewBlock 1 *
NRoots 1
refs CAS(2,2) end
end
NewBlock 3 *
NRoots 1
refs CAS(2,2) end
end
end
This gives the result:
Method S-T gap

MR-CI+Q : 98 cm-1
MR-CI : 93 cm-1
MR-ACPF : 98 cm-1
MR-ACPF2 : 98 cm-1
MR-ACPF2a: 97 cm-1
MR-AQCC : 95 cm-1
SORCI : 131 cm-1
MR-DDCI2 : 85 cm-1
MR-DDCI3 : 130 cm-1
All these methods give good results with SORCI leading to a somewhat larger error than the others. The
(difference dedicated CI) DDCI2 method slightly underestimates the coupling which is characteristic of
this method. It is nice in a way that DDCI3 gives the same result as SORCI since SORCI is supposed to
approximate the DDCI3 (or better the IDDCI3) result which it obviously does.
This splitting can also be studied using broken symmetry HF and DFT methods as explained elsewhere in
this manual:
Method S-T gap

UHF : 70 cm-1
B3LYP/G : 240 cm-1
BP86 : 354 cm-1
PW91 : 234 cm-1
PBE : 234 cm-1
PBE0 : 162 cm-1
RPBE : 242 cm-1
This confirms the usual notions; UHF underestimates the coupling and DFT overestimates it, less so for
hybrid functionals than for GGAs. The BP86 is worse than PW91 or PBE. The PBE0 hybrid may be the
best of the DFT methods. For some reason most of the DFT methods give the best results if the BS state is
simply taken as an approximation for the true open-shell singlet. This is, in our opinion, not backed up by
theory but has been observed by other authors too.
Now let us study the dependence on Tsel as this is supposed to be critical (we use the DDCI3 method):
Tsel S-T gap

1e-04 121
1e-05 128
1e-06 132
1e-07 131
1e-08 131
1e-10 131
1e-12 131
0 131
The convergence is excellent once AllSingles are included.
8.10.8 Potential Energy Surfaces
Another situation where multireference approaches are necessary is when bond breaking is studied and one
wants to calculate a full potential energy surface. Say we want to compute the potential energy surface of the
CH molecule. First we have to figure out which states to include. Hence, let us first determine a significant
number of roots for the full valence CASSCF reference state (we use a small basis set in order to make the
job fast).
! bn-ANO-dzp VeryTightSCF NoPop Conv
%casscf nel 5
norb 5
nroots 2
mult 2
end
%mrci CIType MRCI

NewBlock 2 *
excitations none
NRoots 15
refs CAS(5,5) end
end
NewBlock 4 *
excitations none
NRoots 15
refs CAS(5,5) end
end
end
* xyz 0 2
C 0 0 0
H 0 0 1.15
*
This yields:
-------------------
TRANSITION ENERGIES
-------------------

0 2 -1 0 0 0.000 0.000 0.0
1 2 -1 1 0 0.000 0.000 0.0
2 4 -1 0 1 14.679 0.399 3221.6
3 2 -1 2 0 126.464 3.441 27755.7
4 2 -1 3 0 126.464 3.441 27755.7
5 2 -1 4 0 132.689 3.611 29121.8
6 2 -1 5 0 164.261 4.470 36051.2
7 2 -1 6 0 305.087 8.302 66958.9
8 2 -1 7 0 305.087 8.302 66958.9
9 4 -1 1 1 328.911 8.950 72187.7
10 4 -1 2 1 452.676 12.318 99350.8
11 4 -1 3 1 452.676 12.318 99350.8
12 2 -1 8 0 460.116 12.520 100983.9
13 2 -1 9 0 463.438 12.611 101712.9
14 2 -1 10 0 463.438 12.611 101712.9

...
Thus, if we want to focus on the low-lying states we should include five doublet and one quartet root. Now
we run a second job with these roots and scan the internuclear distance.
! ano-pVDZ VeryTightSCF NoPop Conv MRCI+Q
%casscf nel 5
norb 5
nroots 5,1
mult 2,4
shiftup 2
end
%paras R = 0.8,2.5,25
end
* xyz 0 2
C 0 0 0
H 0 0 {R}
*
The surfaces obtained in this run are shown in 8.17. You can nicely see the crossing of the 2 and 2 states
fairly close to the equilibrium distance and also the merging of the 4 state with 2 and 2 towards the
asymptote that where C-H dossciates in a neutral C-atom in its 3 P ground state and a neutral hydrogen
atom in its 2 S ground state. You can observe that once AllSingles is set to true (the default), the default
settings of the MRCI module yield fairly smooth potential energy surfaces.
In many cases one will focus on the region around the minimum where the surface is nearly quadratic. In this
case one can still perform a few (2, 3, 5, . . . ) point polynomial fitting from which the important parameters
can be determined. The numerical accuracy and the behavior with respect to Tsel has to be studied in these
cases since the selection produces some noise in the procedure. We illustrate this with a calculation on the
HF molecule:
! ano-pVDZ VeryTightSCF NoPop Conv MRCI+Q
%paras R = 0.85,1.1,7
end
%casscf nel 8
norb 5
nroots 1 mult 1
shiftup 2.5 shiftdn 2.5 switchstep nr gtol 1e-5
Figure 8.17: Potential energy surfaces for some low-lying states of CH using the MRCI+Q method
end
%mrci tsel 1e-8

tpre 1e-5
end
* xyz 0 1
F 0 0 0
H 0 0 {R}
*
The output contains the result of a Morse fit:
Morse-Fit Results:
Re = 0.93014 Angstroem
we = 4111.2 cm**-1
wexe = 79.5 cm**-1
Which may be compared with the CCSD(T) values calculated with the same basis set:
Morse-Fit Results:
Re = 0.92246 Angstroem
we = 4209.8 cm**-1
wexe = 97.6 cm**-1
The agreement between MRCI+Q and CCSD(T) results is fairly good.
8.10.9 Multireference Systems - Ozone
The ozone molecule is a rather classical multireference system due to its diradical character. Let us look at
the three highest occupied and lowest unoccupied MO (the next occupied MO is some 6 eV lower in energy
and the next virtual MO some 10 eV higher in energy):
(a) MO-9 (b) MO-10 (c) MO 11(HOMO) (d) MO 12(LUMO)
Figure 8.18: Frontier MOs of the Ozone Molecule.
These MOs are two lone pairs which are high in energy and then the symmetric and antisymmetric
combinations of the oxygen lone pairs. In particular, the LUMO is low lying and will lead to strong
correlation effects since the (HOMO)2 (LUMO)2 excitation will show up with a large coefficient. Physically
speaking this is testimony of the large diradical character of this molecule which is roughly represented by the
structure O-O-O. Thus, the minimal active space to treat this molecule correctly is a CAS(2,2) space which
includes the HOMO and the LUMO. We illustrate the calculation by looking at the RHF, MP2 MRACPF
calculations of the two-dimensional potential energy surface along the OO bond distance and the O-O-O
angle (experimental values are 1.2717 A and 116.78 ).
! ano-pVDZ VeryTightSCF NoPop MRCI+Q Conv
%paras R = 1.20,1.40,21
Theta = 100,150,21
end
%casscf nel 2
norb 2
mult 1
nroots 1
end
%mrci tsel 1e-8

tpre 1e-5
end
* int 0 1
O 0 0 0 0 0 0
O 1 0 0 {R} 0 0
O 1 2 0 {R} {Theta} 0
*
This is a slightly lengthy calculation due to the 441 energy evaluations required. RHF does not find any
meaningful minimum within the range of examined geometries. MP2 is much better and comes close to the
desired minimum but underestimates the OO distance by some 0.03 A. CCSD(T) gives a very good angle
but a OO distance that is too long. In fact, the largest doubles amplitude is 0.2 in these calculations (the
HOMOLUMO double excitation) which indicates a near degeneracy calculation that even CCSD(T) has
problems to deal with. Already the CAS(2,2) calculation is in qualitative agreement with experiment and the
MRCI+Q calculation then gives almost perfect agreement.
The difference between the CCSD(T) and MRCI+Q surfaces shows that the CCSD(T) is a bit lower than
the MRCI+Q one suggesting that it treats more correlation. However, CCSD(T) does it in an unbalanced
way. The MRCI calculation employs single and double excitations on top of the HOMO-LUMO double
excitation, which results in triples and quadruples that apparently play an important role in balancing the
MR calculation. These excitations are treated to all orders explicitly in the MRCI calculation but only
approximately (quadruples as simultaneous pair excitations and triples perturbatively) in the coupled-cluster
approach. Thus, despite the considerable robustness of CC theory in electronically difficult situations it is
not applicable to genuine multireference problems.
This is a nice result despite the too small basis set used and shows how important it can be to go to a
multireference treatment with a physically reasonable active space (even if is only 2 2) in order to get
qualitatively and quantitatively correct results.
8.10.10 Size Consistency
Finally, we want to study the size consistency errors of the methods. For this we study two non-interacting
HF molecules at the single reference level and compare to the energy of a single HF molecule. This should
give a reasonably fair idea of the typical performance of each method (energies in Eh)15 :
E(HF) E(HF+HF) |Difference|

CISD+Q -100.138 475 -200.273 599 0.00335
ACPF -100.137 050 -200.274 010 0.00000
ACPF2 -100.136 913 -200.273 823 0.00000
15
Most of these numbers were obtained with a slightly older version but will not change too much in the present
version.
(a) RHF (b) CASSCF(2,2)
(c) MP2 (d) CCSD(T)
(e) MRCI+Q (f) Difference CCSD(T)/MRCI+Q
Figure 8.19: 2D potential energy surface for the O3 molecule calculated with different methods.
AQCC -100.135 059 -200.269 792 0.00032
The results are roughly as expected CISD+Q has a relatively large error, ACPF and ACPF/2 are perfect
for this type of example; AQCC is not expected to be size consistent and is (only) about a factor of 10 better
than CISD+Q in this respect. CEPA-0 is also size consistent.
8.10.11 Efficient MR-MP2 Calculations for Larger Molecules
Uncontracted MR-MP2 approaches are nowadays outdated. They are much more expensive than internally
contracted e.g. the NEVPT2 method described in section 9.12. Moreover, MR-MP2 is prone to intruder
states, which is a major obstacle for practical applications. For historical reasons, this section is dedicated to
the traditional MR-MP2 approach that is available since version 2.7.0 ORCA. The implementation avoids
the full integral transformation for MR-MP2 which leads to significant savings in terms of time and memory.
Thus, relatively large RI-MR-MP2 calculations can be done with fairly high efficiency. However, the program
still uses an uncontracted first order wavefunction which means that for very large reference space, the
calculations still become untractable.
Consider for example the rotation of the stilbene molecule around the central double bond
Figure 8.20: Rotation of stilbene around the central double bond using a CASSCF(2,2) reference
and correlating the reference with MR-MP2.
The input for this calculation is shown below. The calculation has more than 500 basis functions and still
runs through in less than one hour per step (CASSCF-MR-MP2). The program takes care of the reduced
number of two-electron integrals relative to the parent MRCI method and hence can be applied to larger
molecules as well. Note that we have taken a JK fitting basis in order to fit the Coulomb and the dynamic
correlation contributions both with sufficient accuracy. Thus, this example demonstrates that MR-MP2
calculations for not too large reference spaces can be done efficiently with ORCA (as a minor detail note that
the calculations were started at a dihedral angle of 90 degrees in order to make sure that the correct two
orbitals are in the active space, namely the central carbon p-orbitals that would make up the pi-bond in the
coplanar structure).
#
# Stilbene rotation using MRMP2
#
! def2-TZVP def2/JK RIJCOSX RI-MRMP2
%casscf nel 2
norb 2
end
%mrci maxmemint 2000

tsel 1e-8
end
%paras DIHED = 90,270, 19

end
* int 0 1
C 0 0 0 0.000000 0.000 0.000
8.11 MR-EOM-CC: Multireference Equation of Motion Coupled-Cluster 213
C 1 0 0 1.343827 0.000 0.000

C 2 1 0 1.490606 125.126 0.000
C 1 2 3 1.489535 125.829 {DIHED}
C 4 1 2 1.400473 118.696 180.000
C 4 1 2 1.400488 122.999 0.000
C 6 4 1 1.395945 120.752 180.000
C 5 4 1 1.394580 121.061 180.000
C 8 5 4 1.392286 120.004 0.000
C 3 2 1 1.400587 118.959 180.000
C 3 2 1 1.401106 122.779 0.000
C 11 3 2 1.395422 120.840 180.001
C 12 11 3 1.392546 120.181 0.000
C 13 12 11 1.392464 119.663 0.000
H 1 2 3 1.099419 118.266 0.000
H 2 1 3 1.100264 118.477 179.999
H 5 4 1 1.102119 119.965 0.000
H 6 4 1 1.100393 121.065 0.000
H 7 6 4 1.102835 119.956 180.000
H 8 5 4 1.102774 119.989 180.000
H 9 8 5 1.102847 120.145 180.000
H 10 3 2 1.102271 120.003 0.000
H 11 3 2 1.100185 121.130 0.000
H 12 11 3 1.103001 119.889 180.000
H 13 12 11 1.102704 120.113 180.000
H 14 13 12 1.102746 119.941 180.000
*
8.11 MR-EOM-CC: Multireference Equation of Motion

Coupled-Cluster
The Multireference Equation of Motion Coupled-Cluster (MR-EOM-CC) methodology [141146] has been
implemented in ORCA. The strength of the MR-EOM-CC methodology lies in its ability to calculate many
excited states from a single state-averaged CASSCF solution, the solution of a single set of amplitudes
and an uncontracted MRCI diagonalization, of the final transformed Hamiltonian, over a small manifold of
excited configurations. Hence, a given MR-EOM calculation involves three steps, performed by three separate
modules in ORCA:
1. a state-averaged CASSCF calculation (CASSCF module),
2. the solution of amplitude equations and the calculation of the elements of the similarity transformed
Hamiltonians (MDCI module) and,
3. the uncontracted MRCI diagonalization of the final similarity transformed Hamiltonian (MRCI module).
The current implementation allows for MR-EOM-T|T -h-v, MR-EOM-T|T |SXD-h-v and MR-EOM-T|T |SXD|U-
h-v calculations. A more detailed description of these methods and the available input parameters will be
given in 9.24. We also note that the theoretical details underlying these methods can be found in [146]. In
section 9.24, we will discuss a strategy for the selection of the state-averaged CAS and other steps for setting
up an MR-EOM calculation in detail. Furthermore, we will discuss how spin-orbit coupling effects can be
included in MR-EOM calculations, a projection scheme to aid with convergence difficulties in the iteration of
the T amplitude equations, an orbital selection scheme to reduce the size of the inactive core and virtual
subspaces in the calculation of excitation energies and a strategy for obtaining nearly size-consistent results
in MR-EOM. The purpose of this section is simply to provide a simple example which illustrates the most
basic usage of the MR-EOM implementation in ORCA.
8.11.1 A Simple MR-EOM Calculation
Let us consider an MR-EOM-T|T |SXD|U-h-v calculation on formaldehyde. An MR-EOM-T|T |SXD|U-h-v

calculation is specified via the MR-EOM keyword along with the specification of a state-averaged CASSCF
calculation (i.e. CASSCF(nel, norb) calculation with the number of roots of each multiplicity to be included
in the state-averaging) and the number of desired roots in each multiplicity block for the final MRCI
diagonalization. We note that the CASSCF module is described in sections 8.1.7 and 9.11 and that a
description of the MRCI module is given in sections 8.10 and 9.23. Here, we have a state-averaged CAS(6,4)
calculation, comprised of 3 singlets and 3 triplets and we request 6 singlet roots and 6 triplet roots in our
final MRCI diagonalization (i.e. the roots to be computed in the MR-EOM-T|T |SXD|U-h-v calculation):
!MR-EOM def2-TZVP VeryTightSCF
%casscf
nel 6
norb 4
mult 1,3
nroots 3,3
end
%mdci
STol 1e-7
end
%mrci
newblock 1 *
nroots 6
refs cas(6,4) end
end
newblock 3 *
nroots 6
refs cas(6,4) end
end
end
* xyz 0 1
H 0.000000 0.934473 -0.588078
H 0.000000 -0.934473 -0.588078
C 0.000000 0.000000 0.000000
O 0.000000 0.000000 1.221104
One can alternatively perform an MR-EOM-T|T -h-v or MR-EOM-T|T |SXD-h-v calculation by replacing
the MR-EOM keyword, in the first line of the input above, by MR-EOM-T|Td or MR-EOM-T|Td|SXD,
respectively. Namely, replacing the first line of the input above with
!MR-EOM-T|Td def2-TZVP VeryTightSCF
runs the MR-EOM-T|T -h-v calculation, while,
!MR-EOM-T|Td|SXD def2-TZVP VeryTightSCF
runs the MR-EOM-T|T |SXD-h-v calculation.
The final MR-CI diagonalization manifold includes 2h1p, 1h1p, 2h, 1h and 1p excitations in MR-EOM-T|T -h-
v calculations, 2h, 1p and 1h excitations in MR-EOM-T|T |SXD-h-v calculations and 1h and 1p excitations in
MR-EOM-T|T |SXD|U-h-v calculations. Note that in the mdci block, we have set the convergence tolerance
(STol) for the residual equations for the amplitudes to 107 , as this default value is overwritten with the
usage of the TightSCF, VeryTightSCF, etc. keywords. It is always important to inspect the values of the
largest T, S (i.e. here, we use S to denote the entire set of S, X and D amplitudes) and U amplitudes. If
there are amplitudes that are large (absolute values > 0.15), the calculated results should be regarded with
suspicion. For the above calculation, we obtain:
--------------------
LARGEST T AMPLITUDES
--------------------
8-> 13 8-> 13 0.060329
4-> 17 4-> 17 0.029905
8-> 9 8-> 9 0.028159
8-> 16 8-> 16 0.027265
6-> 20 6-> 20 0.025885
8-> 21 8-> 21 0.025307
4-> 16 4-> 16 0.024802
8-> 12 8-> 12 0.023915
5-> 18 5-> 18 0.023552
8-> 23 8-> 23 0.023384
3-> 16 3-> 16 0.023182
7-> 19 7-> 19 0.023044
8-> 13 4-> 11 0.022009
3-> 19 3-> 19 0.021987
8-> 9 8-> 16 0.021230
8-> 16 8-> 9 0.021230
for the T-amplitudes,

--------------------
LARGEST S AMPLITUDES
--------------------
4-> 8 8-> 11 0.074044
3-> 8 8-> 9 0.064884
4-> 5 5-> 11 0.045476
3-> 8 8-> 16 0.042656
4-> 7 7-> 11 0.042594
4-> 5 5-> 17 0.042074
4-> 5 8-> 11 0.039960
4-> 8 8-> 17 0.037531
3-> 5 8-> 9 0.035908
4-> 7 7-> 17 0.035764
2-> 6 6-> 19 0.034146
3-> 5 5-> 10 0.033339
2-> 6 6-> 10 0.032690
4-> 6 6-> 11 0.032177
8-> 8 3-> 16 0.031774
2-> 7 7-> 22 0.031238
for the S-amplitudes and,
--------------------
LARGEST U AMPLITUDES
--------------------
3-> 8 3-> 8 0.026128
3-> 8 3-> 5 0.007682
2-> 8 2-> 8 0.006182
3-> 8 2-> 5 0.006154
2-> 8 3-> 5 0.004954
3-> 5 3-> 5 0.004677
4-> 8 4-> 8 0.003988
2-> 8 3-> 8 0.002041
3-> 8 2-> 8 0.002041
2-> 8 2-> 5 0.001818
4-> 8 4-> 5 0.001173
2-> 5 2-> 5 0.001107
4-> 5 4-> 5 0.000714
3-> 7 3-> 7 0.000607
3-> 6 3-> 6 0.000521
3-> 5 2-> 5 0.000365
for the U-amplitudes. Hence, one can see that there are no unusually large amplitudes for this particular
calculation. We note that there can be convergence issues with the T amplitude iterations and
that in such cases, the flag:
! RHF TZVP
%cis nroots 1
end
* int 0 1
C 0 0 0 0.000000 0.000 0.000

O 1 0 0 1.200371 0.000 0.000
H 1 2 0 1.107372 121.941 0.000
H 1 2 3 1.107372 121.941 180.000
*
should be added to the %mdci block. The convergence issues are caused by the presence of nearly singular
T2 amplitudes and setting the DoSingularPT flag to true activates a procedure which projects out the
offending amplitudes (in each iteration) and replaces them by suitable perturbative amplitudes. For more
information, see the examples in section 9.24.3.
After the computation of the amplitudes and the elements of the similarity transformed Hamiltonians, within
the MDCI module, the calculation enters the MRCI module. For a complete, step by step description of
the output of an MRCI calculation, we refer the reader to the example described in section 8.10.2. Let us
first focus on the results for the singlet states (CI-BLOCK 1). Following the convergence of the Davidson
diagonalization (default) or DIIS procedure, the following results of the MRCI calculation for the singlet
states are printed:
----------
CI-RESULTS
----------


0.0137 : h---h---[0222]
0.0756 : h---h---[1221]
0.8879 : h---h---[2220]
0.0039 : h---h---[1122]
0.9726 : h---h---[2121]
0.0071 : h---h 4[1222]
0.0085 : h---h 4[2221]
0.0044 : h---h---[1212]
0.9730 : h---h---[2211]
0.0063 : h---h 3[1222]
0.0041 : h---h 3[2221]
0.7459 : h---h---[1221]
0.0807 : h---h---[2022]
0.0533 : h---h---[2220]
0.0228 : h---h 4[2122]
0.0034 : h---h---[1220]p13
0.0072 : h---h---[1220]p18
0.0236 : h---h---[2120]p11
0.0148 : h---h---[2120]p14
0.0069 : h---h---[2120]p17
0.0056 : h---h---[2120]p20
0.0098 : h---h---[2210]p19
0.0045 : h---h---[0122]p9
0.0089 : h---h---[1121]p9
0.9333 : h---h---[2120]p9
0.0243 : h---h---[2120]p10
0.0080 : h---h---[2120]p12
0.0113 : h---h---[2120]p16
0.0061 : h---h---[0222]
0.0918 : h---h---[1221]
0.5785 : h---h---[2022]
0.0048 : h---h---[2202]
0.0047 : h---h---[2220]
0.2904 : h---h 4[2122]
0.0045 : h---h---[2021]p13
For each state, the total energy is given in Eh , the weight of the reference configurations (RefWeight) in the
given state, is provided and the energy differences from the lowest lying state are given in eV and cm1 . Also,
in each case, the weights and a description of the configurations which contribute most strongly to the given
state are also provided. See section 8.10.2 for a discussion of the notation that is used for the description
of the various configurations. In order to avoid confusion, we note that in the literature concerning the
MR-EOM methodology [142148], the term %active is used to denote the reference weight multiplied by
100%. In general, RefWeight should be larger than 0.9, such that the states are dominated by reference space
configurations. This criterion is satisfied for the first three states and the reference weight of the fourth state is
sufficiently close to 0.9. However, the reference weights of the two higher lying states (especially
state 4) are too small and these states should be discarded as the resulting energies will be
inaccurate (i.e. states with significant contributions from configurations outside the reference
space cannot be treated accurately).
In the case of the triplet states (CI-BLOCK 2), we obtain the following results:
----------
CI-RESULTS
----------


0.9691 : h---h---[2121]
0.0079 : h---h 4[1222]
0.0115 : h---h 4[2221]
0.9941 : h---h---[1221]
0.9786 : h---h---[2211]
0.0050 : h---h 3[1222]
0.0044 : h---h---[0122]p9
0.0084 : h---h---[1121]p9
0.9419 : h---h---[2120]p9
0.0131 : h---h---[2120]p10
0.0043 : h---h---[2120]p12
0.0173 : h---h---[2120]p16
0.3862 : h---h---[1122]
0.0154 : h---h---[2121]
0.1721 : h---h 4[1222]
0.4100 : h---h 4[2221]
0.0045 : h---h---[2120]p13
0.0089 : h---h---[0122]p10
0.0030 : h---h---[1121]p10
0.0120 : h---h---[2120]p9
0.9407 : h---h---[2120]p10
0.0105 : h---h---[2120]p16
0.0112 : h---h---[2120]p19
0.0030 : h---h---[2120]p22
Here, we see that the first three states have reference weights which are greater than 0.9, while the reference
weights of the final three states are well below 0.9. Hence, the latter three states should be discarded from
any meaningful analysis.
Following the printing of the CI results for the final CI block, the states are ordered according to increasing
energy and the vertical transition energies are printed:
-------------------
TRANSITION ENERGIES
-------------------

0 1 -1 0 0 0.000 0.000 0.0
1 3 -1 0 1 130.526 3.552 28647.1
2 1 -1 1 0 144.500 3.932 31714.2
3 3 -1 1 1 214.636 5.841 47107.1
4 3 -1 2 1 306.218 8.333 67207.1
5 1 -1 2 0 333.318 9.070 73154.8
6 1 -1 3 0 357.507 9.728 78463.7
7 3 -1 3 1 382.060 10.396 83852.6
8 1 -1 4 0 390.217 10.618 85642.8
9 1 -1 5 0 392.312 10.675 86102.4
10 3 -1 4 1 395.795 10.770 86867.0
11 3 -1 5 1 410.884 11.181 90178.5
Furthermore, following the generation of the (approximate) densities, the absorption and CD spectra are
printed:
------------------------------------------------------------------------------------------
ABSORPTION SPECTRUM
------------------------------------------------------------------------------------------
(cm-1) (nm) (D**2) (D) (D) (D)
------------------------------------------------------------------------------------------
0( 0)-> 1( 0) 1 31714.2 315.3 0.000000000 0.00000 0.00000 -0.00000 0.00000
0( 0)-> 2( 0) 1 73154.8 136.7 0.002133136 0.06192 -0.24884 -0.00000 -0.00000
0( 0)-> 3( 0) 1 78463.7 127.4 0.157692550 4.26771 -0.00000 -0.00000 2.06584
0( 0)-> 4( 0) 1 85642.8 116.8 0.025407931 0.62999 0.00000 -0.79372 0.00000
0( 0)-> 5( 0) 1 86102.4 116.1 0.024717322 0.60959 0.00000 0.00000 0.78076
------------------------------------------------------------------------------
CD SPECTRUM
------------------------------------------------------------------------------
States Energy Wavelength R*T RX RY RZ
(cm-1) (nm) (1e40*sgs) (au) (au) (au)
------------------------------------------------------------------------------
0( 0)-> 1( 0) 1 31714.2 315.3 -0.00000 -0.00000 -0.00000 -1.12539
0( 0)-> 2( 0) 1 73154.8 136.7 0.00000 -0.00000 -1.48989 -0.00000
0( 0)-> 3( 0) 1 78463.7 127.4 0.00000 -0.00000 0.00000 0.00000
0( 0)-> 4( 0) 1 85642.8 116.8 -0.00000 -0.71799 0.00000 0.00000
0( 0)-> 5( 0) 1 86102.4 116.1 -0.00000 0.00000 -0.00000 -0.00000
WARNINGS:
It is important to note that the transition moments and oscillator strengths (and state dipole moments)
have been blindly computed by the MRCI module and currently, no effort has been made to include
the effects of the various similarity transformations in the evaluation of these quantities. Hence these
quantities are only approximate and should only be used as a qualitative aid to determine which states
are dipole allowed or forbidden. Furthermore, since the calculated densities are approximate, so are
the results of the population analysis that are printed before the absorption and CD spectra.
While both the CASSCF and MRCI modules can make use of spatial point-group symmetry to some
extent, the MR-EOM implementation is currently limited to calculations in C1 symmetry.
8.11.2 Capabilities
The MR-EOM methodology can be used to calculate a desired number of states for both closed- and
open-shell systems from a single state-averaged CASSCF solution. Currently, the approach is is limited
to serial calculations and to smaller systems in smaller active spaces. One should be aware that in the
most cost effective MR-EOM-T|T |SXD|U-h-v approach (i.e. smallest diagonalization manifold), an MRCI
diagonalization is performed over all 1h and 1p excited configurations out of the CAS, which will inevitably
limit the size of the initial CAS which can be used. We have also implemented an orbital selection scheme
which can be used to reduce the size of the inactive core and virtual subspaces in the calculation of excitation
energies and this can be employed to extend the applicability of the approach to larger systems. The current
implementation can also be used in conjunction with the spin-orbit coupling submodule (9.23.1) of the MRCI
module to calculate spin-orbit coupling effects in MR-EOM calculations to first-order. These and other
features of the current implementation will be discussed in 9.24.
8.12 Solvation 221
8.12 Solvation
Several implicit solvation models are implemented in ORCA. A completely integrated implementation of the
conductor-like polarizable continuum model (C-PCM) offers a range of options and has been implemented in
all parts of ORCA. The following calculations can be used to carry out calculations in a polarizable continuum
using a realistic Van-der-Waals cavity:
Energies of molecules in solution with a finite dielectric constant using any HF or DFT method.
Optimization of molecular structures in solution using any HF or DFT method using analytic gradients.
Calculation of solvent effects on response properties like polarizabilities through coupled-perturbed

SCF theory. For magnetic response properties such as the g-tensor the C-PCM response vanishes.
Calculations of solvent shifts on transition energies using the time-dependent DFT or CIS method.
Here one needs to know the refractive index of the solvent in addition to the dielectric constant.
First order perturbation estimate of solvent effects on state and transition energies in multireference
perturbation and configuration-interaction calculations.
As a simple example let us look at the solvent effect on the transition energy of the n transition in
formaldehyde. We first do a normal CIS calculation:
! RHF TZVP
%cis nroots 1
end
* int 0 1
C 0 0 0 0.000000 0.000 0.000
O 1 0 0 1.200371 0.000 0.000
H 1 2 0 1.107372 121.941 0.000
H 1 2 3 1.107372 121.941 180.000
*
yielding a transition energy of 4.582 eV. Now we repeat the same calculation but with the CPCM model
enabled (which is fairly simple; nothing except the dielectric constant and the refractive index needs to be
input; there are technical parameters which can be defined by the user but this is not necessary in most
applications; all modules will automatically recognize the presence of the CPCM terms automatically).
! RHF TZVP
%cpcm epsilon 80
refrac 1.33
end
%cis nroots 1
end
* int 0 1
C 0 0 0 0.000000 0.000 0.000
O 1 0 0 1.200371 0.000 0.000
H 1 2 0 1.107372 121.941 0.000
H 1 2 3 1.107372 121.941 180.000
*
This calculation yields:
-------------------------
CALCULATED SOLVENT SHIFTS
-------------------------
State Shift(Eh) Shift(eV) Shift(cm**-1) Shift(nm) ECI(eV) ECI+SHIFT(eV)

-------------------------------------------------------------------
0: -0.0026760 -0.073 -587.3 3.7 4.948 4.875
Note that there are two different types of shift: the first is the difference between the CIS calculation in
the gas phase and the one using the solvated orbitals. This is the difference 4.984 4.582 eV = 0.402 eV.
This term is called the slow term and represents the interaction of the difference density with the frozen
screening charges of the ground state. The second term depends on the instantaneous polarization and is
called the fast term. It is given by the extra 0.073 eV printed above thus yielding a total solvent shift
of 0.402 0.073 eV = 0.329 eV and a final estimate of the transition energy of 4.875 eV in solution (this
is clearly not an accurate value since it is too high as is always found with CIS; the calculated shift is also
somewhat higher compared to accurate MR-CI calculations which give 0.25 eV but it is in the right ballpark;
larger basis sets yield also better results here).
In addition, the Minnesota SMD solvation model is implemented in ORCA. See sections 9.27 and 9.27.3 for
further details on the available solvation models and how to use them.
8.13 Calculation of Properties
8.13.1 Population Analysis and Related Things
Atomic populations and the like are not really a molecular property since they are not observable. They
are nevertheless highly useful for chemical interpretation purposes. ORCA lets you obtain very detailed
information about the calculated molecular orbitals. Mulliken, Lowdin and Mayer population analysis can be
performed and many useful details can be printed. However, it is also easy to get lost in the output from
such a calculation since you may not be interested in all these details. In this case ORCA lets you turn most
features off. The default is to perform a rather large amount of population analysis.
8.13 Calculation of Properties 223
! HF SVP Mulliken Loewdin Mayer ReducedPOP
* xyz 0 1
C 0 0 0
O 0 0 1.13
*
The Mulliken, Lowdin and Mayer analysis tools should be self-explanatory. If you choose ReducedPOP you
will get a reduced orbital population where the percentage contributions per basis function type on each
atom are listed. This is highly useful in figuring out the character of the MOs. You can, however, also
request a printout of the MO coefficients themselves via the output block (section 9.30) or using the keyword
PrintMOs
In many cases it is not so interesting to look at the MO coefficients but you want to get a full three dimensional
picture of MOs, electron densities and spin densities. This is relatively easily accomplished with ORCA
through, among other visualization programs, the interface to the gOpenMol and Molekel packages (see
section 9.31 for details).
The following example:
# test populations
! HF SVP XYZFile
%plots Format gOpenMol_bin
MO("CO-4.plt",4,0);
MO("CO-8.plt",8,0);
end
* xyz 0 1
C 0 0 0
O 0 0 1.13
*
produces (after running it through gOpenMol, section 9.31.2) the following output:
Figure 8.21: The and orbitals of the CO molecule obtained from the interface of ORCA to
gOpenMol.
which are the textbook like and orbitals of CO respectively. The format gOpenMol bin is the most easy
to use. The alternative format gOpenMol ascii would require you to use the gOpenMol conversion utitlity.
You can also plot spin densities, electron densities and natural orbitals. See section 9.31.2 for full details.
The command MO("CO-4.plt",4,0); is to be understood as follows: there is an MO to be evaluated on a
grid and the output is stored in the file CI-4.plt. It is MO four of operator 0. Operator zero is the closed-shell
RHF operator. For UHF wavefunctions operator 0 is that for spin-up and operator 1 that for spin-down. For
ROHF you should also use operator 0. There are also some alternative output formats including simple ascii
files that you can convert yourself to the desired format.
In order to use the interface to Molekel you have to choose the format Cube or Gaussian Cube which can be
read directly by molekel. Since the cube files are ASCII files you can also transfer them between platforms.
# test populations
! HF SVP XYZFile
%plots Format Cube
MO("CO-4.cube",4,0);
MO("CO-8.cube",8,0);
end
* xyz 0 1
C 0 0 0
O 0 0 1.13
*
You can now start Molekel and load (via a right mouse click) the XYZ file (or also directly the .cube file).
Then go to the surface menu, select gaussian-cube format and load the surface. For orbitals click the both
signs button and select a countour value in the cutoff field. The click create surface. The colour schemes
etc. can be adjusted at will try it! Its easy and produces nice pictures. Create files via the snapshot
feature of Molekel. Other programs can certainly also deal with Gaussian-Cube files. If you know about
another nice freeware program please let me know!16
16
The Molekel developers ask for the following citation please do as they ask:
MOLEKEL 4.2, P. Flukiger, H.P. L uthi, S. Portmann, J. Weber, Swiss Center for Scientific Computing, Manno
Figure 8.22: The and -MOs of CO as visualized by Molekel.
Another thing that may in some situations be quite helpful is the visualization of the electronic structure
in terms of localized molecular orbitals. As unitary transformations among the occupied orbitals do not
change the total wavefunction such transformations can be applied to the canonical SCF orbitals with no
change of the physical content of the SCF wavefunction. The localized orbitals correspond more closely to the
pictures of orbitals that chemists often enjoy to think about. Localized orbitals according to the Pipek-Mezey
(population-localization) scheme are quite easy to compute. For example, the following run reproduces the
calculations reported by Pipek and Mezey in their original paper for the N2 O4 molecule. In the output you
will find that the localized set of MOs consists of 6 core like orbitals (one for each N and one for each O), two
distinct lone pairs on each oxygen, a - and a -bonding orbital for each N-O bond and one N-N -bonding
orbital which corresponds in a nice way to the dominant resonance structure that one would draw for this
molecule. You will also find a file with the extension .loc in the directory where you run the calculation.
This is a standard GBW file that you can use for plotting or as input for another calculation (warning! The
localized orbitals have no well defined orbital energy. If you do use them as input for another calculation use
GuessMode=CMatrix in the [SCF] block).
#-----------------------------------------
# Localized MOs for the N2O4 molecule
#-----------------------------------------
! HF STO-3G Bohrs
%loc
LocMet PipekMezey # localization method. Choices:
# PipekMezey (=PM)
# FosterBoys (=FB)
T_Core -1000 # cutoff for core orbitals
Tol 1e-8 # conv. Tolerance (default=1e-6)
MaxIter 20 # max. no of iterations (def. 128)
end
(Switzerland), 2000-2006.
S. Portmann, H.P. L uthi. MOLEKEL: An Interactive Molecular Graphics Tool. CHIMIA (2000), 54, 766-770.
The program appears to be maintained by Ugo Varetto at this time.
* xyz 0 1
N 0.000000 -1.653532 0.000000
N 0.000000 1.653532 0.000000
O -2.050381 -2.530377 0.000000
O 2.050381 -2.530377 0.000000
O -2.050381 2.530377 0.000000
O 2.050381 2.530377 0.000000
*
If you have access to a version of the gennbo program from Weinholds group17 you can also request natural
population analysis and natural bond orbital analysis. The interface is very elementary and is invoked through
the keywords NPA and NBO respectively
# -----------------------------------------
# Test the interface to the gennbo program
# ----------------------------------------------
! HF SVP NPA XYZFile
* xyz 0 1
C 0 0 0
O 0 0 1.13
*
If you choose simple NPA then you will only obtain a natural population analysis. Choosing instead NBO the
natural bond orbital analysis will also be carried out. The program will leave a file jobname.47 on disk. This
is a valid input file for the gennbo program which can be edited to use all of the features of the gennbo
program in the stand-alone mode. Please refer to the NBO manual for further details.
8.13.2 Absorption and Fluorescence Bandshapes
Bandshape calculations are nontrivial but can be achieved with ORCA using the procedures described
in section 9.25. Starting from version 2.80, analytical TD-DFT gradients are available which make these
calculations quite fast and applicable without expert knowledge to larger molecules.
In a nutshell, let us look into the H2 CO molecule. First we generate some Hessian (e.g. BP86/SV(P)). Then
we run the job that makes the input for the orca asa program. For example, let us calculate the five lowest
excited states:
17
Information about the NBO program can be found at http://nbo6.chem.wisc.edu
#
! aug-cc-pVDZ BHandHLYP TightSCF NMGrad
%tddft nroots 5
end
# this is ASA specific input

%rr states 1,2,3,4,5
HessName "Test-ASA-H2CO-freq.hess"
ASAInput true
end
* int 0 1
C 0 0 0 0 0 0
O 1 0 0 1.2 0 0
H 1 2 0 1.1 120 0
H 1 2 3 1.1 120 180
*
NOTE
Functionals with somewhat more HF exchange produce beter results and are not as prone to ghost
states as GGA functionals unfortunately are!
Calculations can be greatly speeded up by the RI or RIJCOSX approximations!
Analytic gradients for the (D) correction and hence for double-hybrid functionals are NOT available
The ORCA run will produce a file Test-ASA-H2CO.asa.inp that is an input file for the program that
generates the various spectra. It is an ASCII file that is very similar in appearance to an ORCA input file:
#
# ASA input
#
%sim model IMDHO
method Heller
AbsRange 25000.0, 100000.0

NAbsPoints 1024
FlRange 25000.0, 100000.0

NFlPoints 1024
RRPRange 5000.0, 100000.0

NRRPPoints 1024
RRSRange 0.0, 4000.0

NRRSPoints 4000
# Excitation energies (cm**-1) for which rR spectra will

# be calculated. Here we choose all allowed transitions

# and the position of the 0-0 band
RRSE 58960, 66884, 66602
# full width half maximum of Raman bands in rR spectra

# (cm**-1):
RRS_FWHM 10.0
AbsScaleMode Ext
FlScaleMode Rel
# RamanOrder=1 means only fundamentals. For 2 combination
# bands and first overtones are also considered, for 3
# one has second overtones etc.
RamanOrder 1
# E0 means the adiabatic excitation energy

# EV would mean the vertical one. sprints vertical
# excitations in the TD-DFT output but for the input into
# the ASA program the adiabatic excitation energies are
# estimated. A rigorous calculation would of course in-
# volve excited state geometry optimization
EnInput E0
CAR 0.800
end
# These are the calculated electronic states and transition moments

# Note that this is in the Franck-Condon approximation and thus
# the transition moments have been calculated vertically
$el_states
5
1 32200.79 100.00 0.00 -0.0000 0.0000 -0.0000
2 58960.05 100.00 0.00 0.0000 -0.4219 0.0000
3 66884.30 100.00 0.00 -0.0000 0.4405 0.0000
4 66602.64 100.00 0.00 -0.5217 -0.0000 0.0000
5 72245.42 100.00 0.00 0.0000 0.0000 0.0000
# These are the calculated vibrational frequencies for the totally

# symmetric modes. These are the only ones tha contribute. They
# correspond to x, H-C-H bending, C=O stretching and C-H stretching
# respectively
$vib_freq_gs
3
1 1462.948534
2 1759.538581
3 2812.815170
# These are the calculated dimensional displacements for all

# electronic states along all of the totally symmetric modes.
$sdnc
3 5
1 2 3 4 5
1 -0.326244 0.241082 -0.132239 0.559635 0.292190
2 -1.356209 0.529823 0.438703 0.416161 0.602301
3 -0.183845 0.418242 0.267520 0.278880 0.231340
Before the orca asa program can be invoked this file must be edited. We turn the NAbsPoints variables and
spectral ranges to the desired values and then invoke orca asa:
orca_asa Test-ASA-H2CO.asa.inp
This produces the output:
******************
* O R C A A S A *
******************
--- A program for analysis of electronic spectra ---
Reading file: Test-ASA-H2CO.asa.inp ... done
**************************************************************
* GENERAL CHARACTERISTICS OF ELECTRONIC SPECTRA *
**************************************************************
--------------------------------------------------------------------------------
State E0 EV fosc Stokes shift Effective Stokes shift
(cm**-1) (cm**-1) (cm**-1) (cm**-1)
--------------------------------------------------------------------------------
1: 30457.24 32200.79 0.000000 0.00 0.00
2: 58424.56 58960.05 0.031879 0.00 0.00
3: 66601.54 66884.30 0.039422 0.00 0.00
4: 66111.80 66602.64 0.055063 0.00 0.00
5: 71788.55 72245.42 0.000000 0.00 0.00
--------------------------------------------------------------------------------------------------
BROADENING PARAMETETRS (cm**-1)
--------------------------------------------------------------------------------------------------
Intrinsic Effective
State -------------------------- --------------------------------------------------------
Sigma FWHM
Gamma Sigma FWHM --------------------------- ---------------------------
0K 77K 298.15K 0K 77K 298.15K
--------------------------------------------------------------------------------------------------
1: 100.00 0.00 200.00 0.00 0.00 0.00 200.00 200.00 200.00
2: 100.00 0.00 200.00 0.00 0.00 0.00 200.00 200.00 200.00
3: 100.00 0.00 200.00 0.00 0.00 0.00 200.00 200.00 200.00
4: 100.00 0.00 200.00 0.00 0.00 0.00 200.00 200.00 200.00
5: 100.00 0.00 200.00 0.00 0.00 0.00 200.00 200.00 200.00
Calculating absorption spectrum ...

The maximum number of grid points ... 5840
Time for absorption ... 9.569 sec (= 0.159 min)
Writing file: Test-ASA-H2CO.asa.abs.dat ... done
Writing file: Test-ASA-H2CO.asa.abs.as.dat ... done
Generating vibrational states up to the 1-th(st) order ... done

Total number of vibrational states ... 3
Calculating rR profiles for all vibrational states up to the 1-th order

State 1 ...
Resonance Raman profile is done
State 2 ...
State 3 ...
Writing file: Test-ASA-H2CO.asa.o1.dat... done
Writing file: Test-ASA-H2CO.asa.o1.info... done
Calculating rR spectra involving vibrational states up to the 1-th(st) order

State 1 ... done
State 2 ... done
State 3 ... done
Writing file: Test-ASA-H2CO.asa.o1.rrs.58960.dat ... done

Writing file: Test-ASA-H2CO.asa.o1.rrs.58960.stk ... done
Writing file: Test-ASA-H2CO.asa.o1.rrs.as.58960.dat ... done
Writing file: Test-ASA-H2CO.asa.o1.rrs.as.58960.stk ... done
Writing file: Test-ASA-H2CO.asa.o1.rrs.all.xyz.dat ... done
TOTAL RUN TIME: 0 days 0 hours 1 minutes 17 seconds 850 msec
The vibrationally resolved absorption spectrum looks like:

The fluorescence spectrum of the lowest energy peak (in this case S2 which is not very realistic but for
illustrative purposes it might be enough):
The Resonance Raman excitation profiles of the three totally symmetric vibrational modes can be obtained
as well:
The dominant enhancement occurs under the main peaks for the C=O stretching vibration which might
not be a big surprise. Higher energy excitations do enhance the C-H vibrations particularly strongly. The
resonance Raman spectra taken at the vertical excitation energies are also calculated:
In this particular example, the dominant mode is the C=O stretching and the spectra look similar for all
excitation wavelength. However, in real life where one has electronically excited states of different nature,
the rR spectra also dramatically change and are then powerful fingerprints of the electronic excitation being
studied even if the vibrational structure of the absorption band is not resolved (which is usually the case
for larger molecules).
This is a cursory example of how to use the orca asa program. It is much more powerful than described in
this section. Please refer to section 9.25 for a full description of features. The orca asa program can also
be interfaced to other electronic structure codes that deliver excited state gradients and can be used to fit
experimental data. It is thus a tool for experimentalists and theoreticians at the same time!
8.13.3 IR/Raman Spectra, Vibrational Modes and Isotope Shifts
8.13.3.1 IR Spectra
IR spectral intensities are calculated automatically in frequency runs. Thus, there is nothing to control by
the user. Consider the following job18 :
! RHF STO-3G TightSCF SmallPrint

! Opt NumFreq
* xyz 0 1
C 0.000000 0.000000 -0.533905
O 0.000000 0.000000 0.682807
H 0.000000 0.926563 -1.129511
H 0.000000 -0.926563 -1.129511
*
which gives you the following output:
-----------
IR SPECTRUM
-----------
Mode freq (cm**-1) T**2 TX TY TZ

-------------------------------------------------------------------
6: 1278.77 6.157280 ( -2.481387 -0.000010 -0.000287)
7: 1395.78 29.682711 ( -0.000003 -5.448182 -0.004815)
8: 1765.08 4.180497 ( 0.000537 -0.022003 2.044508)
9: 2100.53 8.550050 ( 0.000080 0.011990 2.924022)
10: 3498.72 1.843422 ( 0.001027 -0.178764 -1.345907)
11: 3646.23 19.778426 ( 0.000035 4.446922 -0.057577)
The Mode indicates the number of the vibration, then the vibrational frequency follows. The value T**2 is
the square of the change of the dipole moment along a given vibrational mode in KM/mole. This number is
18
This and the following jobs in this section are of very poor quality due to the insufficient quality STO-3G basis set
(see section 10.1). However, these jobs execute very rapidly and are therefore adequate for demonstration purposes.
directly proportional to the intensity of a given fundamental in an IR spectrum and is what is plotted by
orca mapspc.
If you want to obtain a plot of the spectrum then call the small utility program like this:
orca_mapspc Test-NumFreq-H2CO.out ir -w50
The options to the program orca mapspc are:
-wvalue : a value for the linewidth (gaussian shape, fwhm)

-x0value : start value of the spectrum in cm**-1
-x1value : end value of the spectrum in cm**-1
-nvalue : number of points to use
You get a file Test-NumFreq-H2CO.out.ir.dat which contains a listing of intensity versus wavenumber
which can be used in any graphics program for plotting. For example:
Figure 8.23: The predicted IR spectrum of the H2 CO molecule using the numerical frequency rou-
tine of ORCA and the tool orca mapspc to create the spectrum.
8.13.3.2 Raman Spectra
In order to predict the Raman spectrum of a compound one has to know the derivatives of the polarizability
with respect to the normal modes. Thus, if a frequency run is combined with a polarizability calculation the
Raman spectrum will be automatically calculated too.
Consider the following example:

! Opt NumFreq
#
# ... turning on the polarizability calculation
# together with NumFreq automatically gives
# the Raman spectrum
#
%elprop Polar 1
end
* xyz 0 1
c 0.000000 0.000000 -0.533905
o 0.000000 0.000000 0.682807
h 0.000000 0.926563 -1.129511
h 0.000000 -0.926563 -1.129511
*
4
The output consists of the Raman activity (in
A /AMU) [149] and the Raman depolarization ratios:
--------------
RAMAN SPECTRUM
--------------
Mode freq (cm**-1) Activity Depolarization

--------------------------------------------------
6: 1278.77 0.007349 0.749649
7: 1395.78 3.020010 0.749997
8: 1765.08 16.366586 0.708084
9: 2100.53 6.696490 0.075444
10: 3498.72 38.650431 0.186962
11: 3646.23 24.528483 0.748312
The polarizability derivatives and the Raman intensities will also be added to the .hess file. This allows the
effect of isotope substitutions on the Raman intensities to be calculated.
As with IR spectra you can get a plot of the Raman spectrum using:
orca_mapspc Test-NumFreq-H2CO.out raman -w50

Figure 8.24: Calculated Raman spectrum for H2 CO at the STO-3G level using the numerical fre-
quency routine of ORCA and the tool orca mapspc to create the spectrum.
NOTE:
The Raman module will only work if the polarizabilities are calculated analytically. This means that
only those wavefunction models for which the analytical derivatives w.r.t. to external fields are available
can be used.
Raman calculations take significantly longer than IR calculations due to the extra effort of calculating
the polarizabilities at all displaced geometries. Since the latter step is computationally as expensive as
the solution of the SCF equations you have to accept an increase in computer time by a factor of 2.
8.13.3.3 Resonance Raman Spectra
Resonance Raman spectra and excitation profiles can be predicted or fitted using the procedures described in
section 9.25.
8.13.3.4 NRVS Spectra
If you happen to have iron in your molecule and you want to calculate the nuclear resonant vibrational
scattering spectrum you simply have to run orca vib program on your .hess file and you will get an output
that can be used together with orca mapspc program for vizualisation.
orca_vib MyJob.hess >MyJob.vib.out

orca_mapspc MyJob.vib.out NRVS
The calculations are usually pretty good in conjunction with DFT frequency calculations. For example,
take the ferric-azide complex from the second reference listed below. As for the calculation of resonance
Raman spectra described in section 9.25 the DFT values are usually excellent starting points for least-square
refinements.
Figure 8.25: Experimental (a, black curve), fitted (a, red) and simulated (b) NRVS spectrum of
the Fe(III)-azide complex obtained at the BP86/TZVP level (T = 20 K). Bar graphs
represent the corresponding intensities of the individual vibrational transitions. The
blue curve represents the fitted spectrum with a background line removed.
Both theory and implementation have been described in detail. [150, 151] Here we illustrate the procedure for
1
getting such plots using a Fe(SH)4 model complex as an example. One first optimizes and computes the
vibrations of the complex given in one step with the following input (calculations are usually pretty good in
conjunction with DFT frequency calculations and the BP86 functional).
! UKS BP86 def2-TZVP def2/J TightSCF SmallPrint

! Opt Freq
*xyz -1 6
Fe -0.115452 0.019090 -0.059506
S -0.115452 1.781846 1.465006
S -0.115452 -1.743665 1.462801
S -1.908178 -0.072782 -1.518702
S 1.560523 0.154286 -1.656664
H 0.410700 2.760449 0.687716
H -0.674147 -2.708278 0.690223

H -2.905212 0.345589 -0.699907
H 2.647892 -0.211681 -0.932926
*
From this calculations we get numerous files from which the Hessian file is of importance here. Now we
run the orca vib program on the .hess file and get an output that can be used further with orca mapspc
program to prepare raw data for visualizations:
orca vib Test-FeIIISH4-NumFreq.hess > Test-FeIIISH4-NumFreq.out

orca mapspc Test-FeIIISH4-NumFreq.out NRVS
The latter command creates a file Test-FeIIISH4-NumFreq.nrvs.dat which can be used directly for visual-
ization. The text-file contains data in xy-format which allows the NRVS intensity (y, in atomic units) to be
plotted as a function of the phonon enrgy (x, in cm1 ).
From the given run we obtain the NRVS plot below in which we compare with the theoretical IR spectrum on
the same scale. NRVS reports the Doppler broadening of the Moessbauer signal due to resonant scattering of
phonons (vibrations) dominated by the Fe nuclei movements. This are a valuable addition to IR spectra
where the corresponding vibrations might have very small intensity.
8.13.3.5 Animation of Vibrational Modes
In order to animate vibrational modes and to create arrow-pictures you have to use the small utility
program orca pltvib. This program uses an ORCA output file and creates a series of files that can be used
together with any visualization program (here: ChemCraft).
For example:
# NAME = Test-NumFreq-H2CO.inp
! Opt NumFreq
*xyz 0 1
C 0.000000 0.000000 -0.533905
O 0.000000 0.000000 0.682807
H 0.000000 0.926563 -1.129511
H 0.000000 -0.926563 -1.129511
*
From this we get vibrations and transition probabilities (Test-NumFreq-H2CO.out)

Figure 8.26: Theoretical IR spectrum with the shapes of vibrations dominating the IR intensity
and NRVS scattering
Mode freq (cm**-1) T**2 TX TY TZ

-------------------------------------------------------------------
6: 1284.36 6.411244 ( -2.532043 -0.000000 -0.000000)
7: 1397.40 29.590615 ( 0.000000 -5.439726 0.000000)
8: 1766.60 4.188394 ( -0.000000 -0.000000 2.046557)
9: 2099.20 8.429963 ( -0.000000 -0.000000 2.903440)
10: 3499.11 1.792263 ( -0.000000 -0.000000 -1.338754)
11: 3645.24 19.984096 ( 0.000000 -4.470357 0.000000)
The Test-NumFreq-H2CO.out file can be directly opened with ChemCraft which allows immediate observation
of any vibrations and preparing plots as shown.
Figure 8.27: Nuclear vibrations for H2 CO with the shape of each vibration and its frequency
indicated
We can infer for this example, that say the vibration 1397 cm1 is a kind of wagging motion of the hydrogen
atoms.
It might be that you can prefer to animate vibrations with the (free) program gOpenMol package; there is a
small utility program orca pltvib. This program uses an ORCA output file and creates a series of files that
can be used together with gOpenMol. You can execute orca pltvib in the following way:
Use:
orca_pltvib Test-NumFreq-H2CO.out [list of vibrations or all]
For example, let us see what the strong mode at 1397 cm1 corresponds to:
orca_pltvib Test-NumFreq-H2CO.out 7
you will get a file Test-NumFreq-H2CO.out.v007.xyz. Then start up the gOpenMol program and read this
file as a Import->coords in Xmol format. After this go to the Trajectory->Main menu and import the
file again (again in Xmol format). Now you are able to animate the vibration. In order to create a printable
picture press Dismiss and then type lulVectorDemo {4 0.1 black} into the gOpenMol command line
window. What you get is:
which indicates that the vibration is a kind of wagging motion of the hydrogens. (I am sure that you can get
nicer arrows with some playing around with gOpenMol). At the gOpenMol homepage you can find a very
nice tutorial to teach you some essential visualization tricks.
Figure 8.28: The 1395 cm1 mode of the H2 CO molecule as obtained from the interface of ORCA
to gOpenMol and the orca pltvib tool to create the animation file.
8.13.3.6 Isotope Shifts
Suppose you have calculated a hessian as in the example discussed above and that you want to predict the
effect of substitution with 18 O. It would be very bad pratice to recalculate the hessian to do this since the
calculation is expensive and the hessian itself is independent of the masses. In this case you can use the small
utility program orca vib. First of all you have to look at the .hess file and edit the masses given there by
hand. For the example given above the .hess file looks like:
$orca_hessian_file
......................
$hessian
12
... the cartesian hessian in Eh/bohr**"
$vibrational_frequencies
12
...the vibrational frequencies (in cm-1) as in the output
$normal_modes
12 12
... the vibrational normal modes in Cartesian displacements
#
# The atoms: label mass x y z
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# Here we have changes 15.999 for oxygen into
# 18.0 in order to see the oxygen 18 effects
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
$atoms
4
C 12.0110 0.000000 0.000000 -1.149571
O 18.0000 -0.000000 -0.000000 1.149695
H 1.0080 -0.000000 1.750696 -2.275041
H 1.0080 -0.000000 -1.750696 -2.275041
$actual_temperature
0.000000
$dipole_derivatives
12
... the dipole derivatives (Cartesian displacements)

#
# The IR spectrum
# wavenumber T**2 TX TY TY
#
$ir_spectrum
12
... the IR intensities
if you now call:
orca_vib Test-NumFreq-H2CO.hess
you get then printed to standard out the IR spectrum, the vibrational frequencies, the modes, etc. Let us
compare the output of this calculation with the original frequency calculation:
H2C16O H2CO18O Shift

6: 1284.36 1282.82 cm**-1 -1.54
7: 1397.40 1391.74 cm**-1 -5.66
8: 1766.60 1751.62 cm**-1 -14.98
9: 2099.20 2061.49 cm**-1 -37.71
10: 3499.11 3499.02 cm**-1 -0.09
11: 3645.24 3645.24 cm**-1 -0.00
The calculated isotope shifts greatly aid in the identification of vibrations, the interpretation of experiments
and in the judgement of the reliability of the calculated vibrational normal modes. A different way of
analyzing these isotope shifts is to plot the two predicted spectra and then subtract them from another. This
will give you derivative shaped peaks with a zero crossing at the position of the isotope sensitive modes.
8.13.4 Thermochemistry
The second thing that you get automatically as the result of a frequency calculation is a thermochemical
analysis based on ideal gas statistical mechanics. This can be used to study heats of formation, dissociation
energies and similar thermochemical properties. To correct for the breakdown of the harmonic oscillator
approximation for low frequencies, entropic contributions to the free energies are computed, by default, using
the Quasi-RRHO approach of Grimme. [152] To switch-off the Quasi-RRHO method, use:
%freq QuasiRRHO false

CutOffFreq 35
end
Where the CutOffFreq parameter controls the cut-off for the low frequencies mode (excluded from the
calculation of the thermochemical properties).
For example let us calculate a number for the oxygen-oygen dissociation energy in the H2 O2 molecule. First
run the following job:
#
# Calculate a value for the O-O bond strength in H2O2
#
! RKS B3LYP TZVP TightSCF SmallPrint Grid4 NoFinalgrid
! Opt NumFreq
! bohrs
* xyz 0 1
O -1.396288 -0.075107 0.052125
O 1.396289 -0.016261 -0.089970
H -1.775703 1.309756 -1.111179
H 1.775687 0.140443 1.711854
*
#
# Now the OH radical job
#
$new_job
! UKS B3LYP TZVP TightSCF SmallPrint Grid4 NoFinalgrid
! Opt NumFreq PModel
! bohrs
* xyz 0 2
O -1.396288 -0.075107 0.052125
H -1.775703 1.309756 -1.111179
*
The first job gives you the following output following the frequency calculation:
--------------------------
THERMOCHEMISTRY AT 298.15K
--------------------------
Temperature ... 298.15 K

Pressure ... 1.00 atm
Total Mass ... 34.01 AMU
Throughout the following assumptions are being made:

(1) The electronic state is orbitally nondegenerate
(2) There are no thermally accessible electronically excited states
(3) Hindered rotations indicated by low frequency modes are not
treated as such but are treated as vibrations and this may
cause some error
(4) All equations used are the standard statistical mechanics
equations for an ideal gas
(5) All vibrations are strictly harmonic
------------
INNER ENERGY
------------
The inner energy is: U= E(el) + E(ZPE) + E(vib) + E(rot) + E(trans)

E(el) - is the total energy from the electronic structure calculation
= E(kin-el) + E(nuc-el) + E(el-el) + E(nuc-nuc)
E(ZPE) - the zero temperature vibrational energy from the frequency calculation
E(vib) - the finite temperature correction to E(ZPE) due to population
of excited vibrational states
E(rot) - is the rotational thermal energy
E(trans)- is the translational thermal energy
Summary of contributions to the inner energy U:

Electronic energy ... -151.54452527 Eh
Zero point energy ... 0.02624039 Eh 16.47 kcal/mol
Thermal vibrational correction ... 0.00044174 Eh 0.28 kcal/mol
Thermal rotational correction ... 0.00141627 Eh 0.89 kcal/mol
Thermal translational correction ... 0.00141627 Eh 0.89 kcal/mol
-----------------------------------------------------------------------
Total thermal energy -151.51501060 Eh
Summary of corrections to the electronic energy:

(perhaps to be used in another calculation)
Total thermal correction 0.00327428 Eh 2.05 kcal/mol
Non-thermal (ZPE) correction 0.02624039 Eh 16.47 kcal/mol
-----------------------------------------------------------------------
Total correction 0.02951468 Eh 18.52 kcal/mol
--------
ENTHALPY
--------
The enthalpy is H = U + kB*T

kB is Boltzmanns constant
Total free energy ... -151.51501060 Eh
Thermal Enthalpy correction ... 0.00094421 Eh 0.59 kcal/mol
-----------------------------------------------------------------------
Total Enthalpy ... -151.51406639 Eh
-------
ENTROPY
-------
The entropy contributions are T*S = T*(S(el)+S(vib)+S(rot)+S(trans))

S(el) - electronic entropy
S(vib) - vibrational entropy
S(rot) - rotational entropy
S(trans)- translational entropy
The entropies will be listed as multiplied by the temperature to get
units of energy
Electronic entropy ... 0.00000000 Eh 0.00 kcal/mol

Vibrational entropy ... 0.00066248 Eh 0.42 kcal/mol
Rotational entropy ... 0.00752687 Eh 4.72 kcal/mol
Translational entropy ... 0.01734394 Eh 10.88 kcal/mol
-----------------------------------------------------------------------
Final entropy term ... 0.02553329 Eh 16.02 kcal/mol
-------------------
GIBBS FREE ENTHALPY
-------------------
The Gibbs free enthalpy is G = H - T*S
Total enthalpy ... -151.51406639 Eh

Total entropy correction ... -0.02553329 Eh -16.02 kcal/mol
-----------------------------------------------------------------------
Final Gibbs free enthalpy ... -151.53959968 Eh
For completeness - the Gibbs free enthalpy minus the electronic energy
G-E(el) ... 0.00492559 Eh 3.09 kcal/mol
And similarly for the OH-radical job.
------------
INNER ENERGY
------------
Summary of contributions to the inner energy U:

Electronic energy ... -75.73288320 Eh
Zero point energy ... 0.00838485 Eh 5.26 kcal/mol
Thermal vibrational correction ... 0.00000000 Eh 0.00 kcal/mol
Thermal rotational correction ... 0.00094418 Eh 0.59 kcal/mol
Thermal translational correction ... 0.00141627 Eh 0.89 kcal/mol
-----------------------------------------------------------------------
Total thermal energy -75.72213790 Eh
Summary of corrections to the electronic energy:

(perhaps to be used in another calculation)
Total thermal correction 0.00236045 Eh 1.48 kcal/mol
Non-thermal (ZPE) correction 0.00838485 Eh 5.26 kcal/mol
-----------------------------------------------------------------------
Total correction 0.01074531 Eh 6.74 kcal/mol
--------
ENTHALPY
--------
The enthalpy is H = U + kB*T

kB is Boltzmanns constant
Total free energy ... -75.72213790 Eh
Thermal Enthalpy correction ... 0.00094421 Eh 0.59 kcal/mol
-----------------------------------------------------------------------
Total Enthalpy ... -75.72119369 Eh
-------
ENTROPY
-------
Electronic entropy ... 0.00065446 Eh 0.41 kcal/mol

Vibrational entropy ... 0.00000000 Eh 0.00 kcal/mol
Rotational entropy ... 0.00256479 Eh 1.61 kcal/mol
Translational entropy ... 0.01636225 Eh 10.27 kcal/mol
-----------------------------------------------------------------------
Final entropy term ... 0.01958150 Eh 12.29 kcal/mol
-------------------
GIBBS FREE ENTHALPY
-------------------
The Gibbs free enthalpy is G = H - T*S
Total enthalpy ... -75.72119369 Eh

Total entropy correction ... -0.01958150 Eh -12.29 kcal/mol
-----------------------------------------------------------------------
Final Gibbs free enthalpy ... -75.74077518 Eh
For completeness - the Gibbs free enthalpy minus the electronic energy
G-E(el) ... -0.00789198 Eh -4.95 kcal/mol
Let us calculate the free energy change for the reaction:
H2O2 -> 2 OH
The individual energy terms are: 2OH(Eh) - H2O(Eh) kcal/mol
Electronic Energy: (-151.465766)-(-151.544525) 49.42
Zero-point Energy: (0.016770)-(0.026240) -5.94
Thermal Correction(translation/rotation): (0.00472)-(0.002832) 1.18
Thermal Enthalpy Correction: (0.001888)-(0.000944) 0.59
Entropy: -(0.039163)-(-0.025533) -8.55
Final G 36.70
Thus, both the zero-point energy and the entropy terms both contribute significantly to the total free energy
change of the reaction. The entropy term is favoring the reaction due to the emergence of new translational
and rotational degrees of freedom. The zero-point correction is also favoring the reaction since the zero-point
vibrational energy of the O-O bond is lost. The thermal correction and the enthalpy correction are both
small.
TIP:
You can run the thermochemistry calculations at several user defined temperatures by providing the
program with a list of temperatures:
%freq Temp 290, 295, 300

end
8.13.5 Anharmonic Analysis
Building upon (analytical) harmonic calculations of the Hessian, it is possible to calculate a semi-quartic force
field as well as higher-order property derivatives. These are required for an anharmonic analysis according to
second-order vibrational perturbation theory. The VPT2 calculation can be initiated either via the simple
input command !VPT2 or via the VPT2 keyword in the %vpt2 block. Finer control can be achieved through
the %VPT2 block, as exemplified in this analysis of water.
# VPT2 Analysis of H2 O
!RHF def2-SVP ExtremeSCF
!VPT2
%vpt2
VPT2 On # do a VPT2 analysis, same as !VPT2 (see above)
AnharmDisp 0.05 # anharmonic displacement factor, default is 0.05
HessianCutoff 1e-12 # cut-off for Hessian matrix elements, default is 1010
PrintLevel 1 # VPT2 print level [1, 2, 3, 4]
MinimiseOrcaPrint True # Minimises the remaining orca output
end
%method
Z_Tol 1e-14
end
* xyz 0 1
O 0.00000000000000 0.06256176106279 0.06256176106280
H 0.00000000000000 -0.06185639479702 0.99929463373422
H 0.00000000000000 0.99929463373424 -0.06185639479703
*
After the analysis, a <basename>.vpt2 file should be present in the working directory. Within that file all the
force field and property derivatives are saved. It is used as an input for the orca vpt2 programme which is
called automatically after the initial displacement calculations. The programme can also be called separately
with the command orca vpt2 <basename>.vpt2.
A few remarks about VPT2 calculations:
A VPT2 starting geometry should always be tightly converged. For small molecules the !TightOPT
option is not good enough. Depending on your structure, you might want to experiment with the TolE,
TolRMSG and TolMaxG keywords of the %geom block.
Similarly, a well converged SCF is required. The use of the ExtremeSCF (or similar) keyword is
recommended.
The CP-SCF equations should be converged to at least 1012 (modified via the Z Tol setting in the
%method block.
Linear molecules are not supported yet
Currently, only HartreeFock, DFT and MP2 calculations are supported. Furthermore, VPT2 calcula-
tions with DFT functionals which do not provide analytical Hessians cannot be carried out.
By default, updated atomic masses are used to generate the semi-quartic force field (see 8.3). The
masses are printed in the <basename>.vpt2 file and can be changed in the main input by adding M
<new mass> behind the coordinate input of each atom.
Currently the programme version is being tested. It provides a vibrational-rotational analysis and thus access
to B 0 , the rotational constants at the ro-vibrational ground state. Furthermore orca vpt2 calculates, the
anharmonic constants rs , the rotational-vibrational coupling constants ri , the zero-point ro-vibrational
energy contributions as well as the harmonic and fundamental transition frequencies r and r .
While the Coriolis-resonance corrections are implemented, the Fermi-resonance corrections are still missing.
Anharmonic corrections to the harmonic intensities will soon follow.
8.13.6 Electrical Properties
A few basic electric properties can be calculated in ORCA although this has never been a focal point of
development. The properties can be accessed straightforwardly through the %elprop block:
! RKS B3LYP SVP SmallPrint TightSCF Grid4
%elprop Dipole true

Quadrupole True
Polar 1 # analytic polarizability through CP-SCF
# possible for SCF runs (HF and DFT)
# = 3: fully numeric
end
* int 0 1
C 0 0 0 0 0 0
H 1 0 0 1.09 109.4712 0
H 1 2 0 1.09 109.4712 0
H 1 2 3 1.09 109.4712 120
H 1 2 3 1.09 109.4712 240
The polarizability is calculated analytically through solution of the coupled-perturbed SCF equations for HF
and DFT runs. For MP2 one can differentiate the analytical dipole moment calculated with relaxed densities.
For other correlation methods only a fully numeric approach is possible.
-------------------------
THE POLARIZABILITY TENSOR
-------------------------
The raw cartesian tensor (atomic units):

12.85800 -0.00001 -0.00000
-0.00001 12.85797 -0.00000
-0.00000 -0.00000 12.85794
diagonalized tensor:
12.85794 12.85797 12.85801
0.00022 -0.24634 0.96918

0.00063 -0.96918 -0.24634
1.00000 0.00067 -0.00005
Isotropic polarizability : 12.85797
As expected the polarizability tensor is isotropic.
The following jobs demonstrate the numeric and analytic calculations of the polarizability:
# ---------------------------------------------
# Numerical calculation of the polarizability
# ---------------------------------------------
! UKS B3LYP/G SVP VeryTightSCF
%elprop Polar 3
EField 1e-5
end
* int 1 2
C 0 0 0 0 0 0
O 1 0 0 1.1105 0 0
*
$new_job
# ---------------------------------------------
# Analytical calculation of the polarizability
# ---------------------------------------------
! UKS B3LYP/G SVP VeryTightSCF
%elprop Polar 1
Tol 1e-7
end
* int 1 2
C 0 0 0 0 0 0
O 1 0 0 1.1105 0 0
*
Here the polarizability of CO+ is calculated twice first numerically using a finite field increment of 105 au
and then analytically using the CP-SCF method. In general the analytical method is much more efficient,
especially for increasing molecular sizes.
At the MP2 level, polarizabilities can currently be calculated analytically in all-electron calculations, but
with frozen core orbitals the dipole moment has got to be differentiated numerically in order to obtain the
polarizability tensor. This will in general require tight SCF converge in order to not get too much numerical
noise in the second derivative. Also, you should experiment with the finite field increment in the numerical
differentiation process. For example consider the following simple job:
! RHF MP2 SVP VeryTightSCF

%elprop Polar 2
EField 1e-4
end
* int 0 1
C 0 0 0 0 0 0
O 1 0 0 1.130 0 0
*
In a similar way, polarizability calculations are possible with CASSCF. For other correlation methods, where
not even response densities are available, only a fully numeric approach (Polar=3) is possible and requires
obnoxiously tight convergence.
Note that polarizability calculations have higher demands on basis sets. A rather nice basis set for this
property is the Sadlej one (see 9.3.1).
8.13.7 NMR Chemical Shifts
NMR chemical shifts at the HF and DFT (standard GGA and hybrid functionals) levels can be obtained
from the EPR/NMR module of ORCA. For the calculation of the NMR shielding tensor the program utilizes
Gauge Including Atomic Orbitals (GIAOs - sometimes also refered to as London orbitals). [153155] In this
approach, field dependet basis functions are introduced, which minimizes the gauge origin dependence and
ensures rapid convergence of the results with the one electron basis set. [156]
The use of the chemical shift module is simple:

# Ethanol - Calculation of the NMR chemical shieldings at the HF/SVP level of theory
! RHF SVP def2/JK Bohrs NMR
* xyz 0 1
C -1.22692181 0.24709455 -0.00000000
C -0.01354839 -0.54677253 0.00000000
H -2.09280406 -0.41333631 0.00000000
H -1.24962478 0.87541936 -0.88916500
H -1.24962478 0.87541936 0.88916500
O 1.09961824 0.30226226 -0.00000000
H 0.00915178 -1.17509696 0.88916500
H 0.00915178 -1.17509696 -0.88916500
H 1.89207683 -0.21621566 0.00000000
*
The output for the shieldings contains detailed information about the para- and diamagnetic contribution,
the orientation of the tensor, the eigenvalues, its isotropic part etc. For each atom, an output block with this
information is given :
--------------
Nucleus 0C :
--------------
Diamagnetic contribution to the shielding tensor (ppm) :

209.678 -10.541 -0.000
-26.566 215.778 0.000
-0.000 0.000 200.386
Paramagnetic contribution to the shielding tensor (ppm):

59.273 18.302 0.000
13.380 6.063 -0.000
0.000 -0.000 -2.770
Total shielding tensor (ppm):

268.951 7.760 -0.000
-13.185 221.841 0.000
0.000 -0.000 197.615
Diagonalized sT*s matrix:
sDSO 200.386 214.435 211.021 iso= 208.614

sPSO -2.770 7.271 58.064 iso= 20.855
--------------- --------------- ---------------
Total 197.615 221.707 269.085 iso= 229.469
Note that all units are given in ppm and the chemical shieldings given are absolute shieldings (see below). At
the end of the atom blocks, a summary is given with the isotropic shieldings and the anisotropy [157] for
each nucleus:
Nucleus Element Isotropic Anisotropy

------- ------- ------------ ------------
0 C 229.469 59.424
1 C 227.615 62.882
2 H 55.997 12.498
3 H 55.445 15.254
4 H 55.445 15.254
5 O 334.100 110.659
6 H 47.326 27.102
7 H 47.326 27.102
8 H 64.238 32.111
Thus, the absolute, isotropic shielding for the 13 C nuclei are predicted to be 229.5 and 227.6 ppm and for
17
O it is 334.1 ppm. While basis set convergence using GIAOs is rapid and smooth, it is still recommended
to do NMR calculations with basis sets including tight exponents. However, TZVPP or QZVP should be
sufficient in most cases. [158, 159]
An important thing to note is that in order to compare to experiment, a standard molecule for the type of
nucleus of interest has to be chosen. In experiment, NMR chemical shifts are usually determined relative to a
standard, for example either CH4 or TMS for proton shifts. Hence, the shieldings for the molecule of interest
and a given standard molecule are calculated, and the relative shieldigs are obtained by subtraction of the
reference value from the computed value. It is of course important that the reference and target calculations
have been done with the same basis set and functional. This also helps to benefit from error cancellation if
the standard is chosen appropriately (one option is even to consider an internal standard - that is to use
for example the 13 C shielding of a methyl group inside the compound of interest as reference).
Let us consider an example - propionic acid (CH3 -CH2 COOH). In databases like the AIST (http://sdbs.
db.aist.go.jp) the 13 C spectrum in CDCl3 can be found. The chemical shifts are given as 1 = 8.9 ppm,
2 = 27.6 ppm, 3 = 181.5 ppm. While intuition already tells us that the carbon of the carboxylic acid group
should be shielded the least and hence shifted to lower fields (larger values), lets look at what calcuations
at the HF, BP86 and B3LYP level of theory using the SVP and the TZVPP basis sets yield:
method 1 2 3
HF/SVP 191.7 176.6 23.7
HF/TZVPP 183.5 167.1 9.7
B86/SVP 181.9 165.8 26.5
B86/TZVPP 174.7 155.5 7.6
B3LYP/SVP 181.8 165.8 22.9
B3LYP/TZVPP 173.9 155.0 2.9
Looking at these results, we can observe several things - first of all, the dramatic effect of using too small
basis sets, which yields differences of more than 10 ppm. Seond, the results obviously change a lot upon
inclusion of electron correlation by DFT and are functional dependent. Last but not least, these values have
nothing in common with the experimental ones (they change in the wrong order), as the calculation yields
absolute shieldings like in the table above, but the experimental ones are relative shifts, in this case relative
to TMS.
In order to obtain the relative shifts, we calculate the shieldings T M S of the standard molecule (TMS
HF/TZVPP: 194.1 ppm, BP86/TZVPP: 184.8 ppm, B3LYP/TZVPP: 184.3 ppm) and by using mol =
ref mol we can evaluate the relative chemical shieldings (in ppm) and directly compare to experiment:
method 1 2 3
HF/TZVPP 10.6 27.0 184.4
B86/TZVPP 10.1 29.3 177.2
B3LYP/TZVPP 10.4 29.3 181.4
Exp. 8.9 27.6 181.5
A few notes on the GIAO implementation in ORCA are in order. The use of GIAOs lead to some fairly
complex molecular one- and two-electron integrals and a number of extra terms on the right hand side of the
coupled-perturbed SCF equations. These contributions can be time consuming to calculate. In the present
ORCA implementation the four-center two-electron GIAO integrals are fairly slow. Hence, we recommend
to only use them for reference type calculations on small molecules. A variety of approximations were
implemented and are currently being tested. The most satisfactory of these approximations is the RI-JK
approximation. Hence, it has presently been made the default. This means, that - if you follow the defaults -
you have to provide an auxiliary basis set, even if the SCF calculation is done without any approximation.
Please note that the scaling of RIJK is the same as in the SCF, e.g. fourth power of the system size with a
small prefactor. Hence, for large molecules, these calculations will be time consuming.
The approximation can be controlled in the eprnmr input block
%eprnmr GIAO_1el = giao_1el_analytic # analytic integration (default)

= giao_1el_numeric # numeric integration
GIAO_2el = giao_2el_analytic # fully analytic handling of 2el terms
= giao_2el_rijonx # Coulomb with RI, exchange exact
= giao_2el_rijcosx # Coulomb with RI, exchange COSX
= giao_2el_rijk # RI for Coulomb and exchange (default)
= giao_2el_cosk # Chain of spheres for Coulomb and exchange
end
It is also emphasized that the user can finely control for which nuclei the shifts are calculated. This works
in exactly the same way as for the hyperfine and quadrupole couplings described in the next section. For
example:
! B3LYP def2-TZVP def2/JK TightSCF
* int 0 1
C 0 0 0 0 0 0
C 1 0 0 1.35 0 0
H 1 2 0 1.1 120 0
H 1 2 3 1.1 120 180
H 2 1 3 1.1 120 0
H 2 1 3 1.1 120 180
*
%eprnmr Nuclei = all C { shift }

Nuclei = all H { shift }
end
8.13.8 Hyperfine and Quadrupole Couplings
Hyperfine and quadrupole couplings can be obtained from the EPR/NMR module of ORCA. Since there
may be several nuclei that you might be interested in the input is relatively sophisticated.
An example how to calculate the hyperfine and field gradient tensors for the CN radical is given below:
! UKS PBE0 DZP SmallPrint TightSCF Grid5

* int 0 2
C 0 0 0 0 0 0
N 1 0 0 1.170 0 0
*
%eprnmr Nuclei = all C { aiso, adip }
Nuclei = all N { aiso, adip, fgrad }
end
In this example the hyperfine tensor is calculated for all carbon atoms and all nitrogen atoms which, of
course, happen to be a single atom in this specific case. In general you can also give the numbers of certain
atoms instead of all in order to avoid undesired calculations. A more subtle point is that you have
to specify the Nuclei statement after the definition of the atomic coordinates or the program
will not figure out what is meant by all.
The output looks like the following. It contains similar detailed information about the individual contributions
to the hyperfine couplings, its orientation, its eigenvalues, the isotropic part and (if requested) also the
quadrupole coupling tensor.
-----------------------------------------
ELECTRIC AND MAGNETIC HYPERFINE STRUCTURE
-----------------------------------------
-----------------------------------------------------------
Nucleus 0C : I= 0.5 P=134.1900 MHz/au**3 Q= 0.0000 barn
-----------------------------------------------------------
Raw HFC matrix (all values in MHz):
696.2146 -0.0000 -0.0000
-0.0000 542.9631 -0.0000
-0.0000 -0.0000 542.9631
A(FC) 594.0469 594.0469 594.0469

A(SD) -51.0838 -51.0838 102.1677
---------- ---------- ----------
A(Tot) 542.9631 542.9631 696.2146 A(iso)= 594.0469
Orientation:
X -0.0000000 -0.0000000 -1.0000000
Y -0.9999951 0.0031335 0.0000000
Z -0.0031335 -0.9999951 0.0000000
-----------------------------------------------------------
Nucleus 1N : I= 1.0 P= 38.5677 MHz/au**3 Q= 0.0193 barn
-----------------------------------------------------------
Raw HFC matrix (all values in MHz):
13.1112 0.0000 0.0000
0.0000 -45.8793 0.0000
0.0000 0.0000 -45.8793
A(FC) -26.2158 -26.2158 -26.2158

A(SD) 39.3270 -19.6635 -19.6635
---------- ---------- ----------
A(Tot) 13.1112 -45.8793 -45.8793 A(iso)= -26.2158
Orientation:
X -1.0000000 0.0000000 0.0000000
Y -0.0000000 -0.9999998 0.0005676
Z -0.0000000 -0.0005676 -0.9999998
Raw EFG matrix (all values in a.u.**-3):

-0.1952 0.0000 0.0000
0.0000 0.0976 0.0000
0.0000 0.0000 0.0976
V(El) 0.6527 0.6527 -1.3054

V(Nuc) -0.5551 -0.5551 1.1102
---------- ---------- ----------
V(Tot) 0.0976 0.0976 -0.1952
Orientation:
X -0.0000001 0.0000001 1.0000000
Y 0.0069379 0.9999759 -0.0000001
Z -0.9999759 0.0069379 -0.0000001
Quadrupole tensor eigenvalues (in MHz;Q= 0.0193 I= 1.0)

e**2qQ = -0.886 MHz
e**2qQ/(4I*(2I-1))= -0.222 MHz
eta = 0.000
NOTE: the diagonal representation of the SH term
I*Q*I = e**2qQ/(4I(2I-1))*[-(1-eta),-(1+eta),2]
Another point for hyperfine calculations is that you should normally use basis sets that have more flexibility
in the core region. In the present example a double-zeta basis set was used. For accurate calculations you
need more flexible basis sets. There are several dedicated basis set for hyperfine calculations: (a) the EPR-II
basis of Barone and co-workers. It is only available for a few light atoms (H, B, C, N, O, F). It is essentially
of double-zeta plus polarization quality with added flexibility in the core region and should give reasonable
results (b) The IGLO-II and IGLO-III bases of Kutzelnigg and co-workers. They are fairly accurate but also
only available for some first and second row elements (c) the CP basis for first row transition metals which is
accurate as well. (d) General purpose HF-limit basis sets are the uncontracted Partridge basis sets. They
will probably be too expensive for routine use but are useful for calibration purposes.
For other elements ORCA does not yet have dedicated default basis sets for this situation it is very likely that
you have to tailor the basis set to your needs. If you use the statement Print[p basis] 2 in the %output
block (or PrintBasis in the simple input line) the program will print the actual basis set in input format
(for the basis block). You can then add or remove primitives, uncontract core bases etc. For example, here is
a printout of the carbon basis DZP in input format:
# Basis set for element : C

NewGTO 6
s 5
1 3623.8613000000 0.0022633312
2 544.0462100000 0.0173452633
3 123.7433800000 0.0860412011
4 34.7632090000 0.3022227208
5 10.9333330000 0.6898436475
s 1
1 3.5744765000 1.0000000000
s 1
1 0.5748324500 1.0000000000
s 1
1 0.1730364000 1.0000000000
p 3
1 9.4432819000 0.0570590790
2 2.0017986000 0.3134587330
3 0.5462971800 0.7599881644
p 1
1 0.1520268400 1.0000000000
d 1
1 0.8000000000 1.0000000000
end;
The s 5, for example, stands for the angular momentum and the number of primitives in the first basis
function. Then there follow five lines that have the number of the primitive, the exponent and the contraction
coefficient (unnormalized) in it. Remember also that when you add very steep functions you must
increase the size of the integration grid if you do DFT calculations! If you do not do that your
results will be inaccurate. You can increase the radial grid size by using IntAcc in the Method block or
for individual atoms (section 9.2.2.2 explains how to do this in detail). In the present example the changes
caused by larger basis sets in the core region and more accurate integration are relatively modest on the
order of 3%, which is, however, still significant if you are a little puristic.
The program can also calculate the spin-orbit coupling contribution to the hyperfine coupling tensor as
described in section 9.28.3.To extract the A tensor from a oligonuclear transition metal complex, the A(iso)
value in the output is to be processed according to the method described in ref. [160].
If also EPR g-tensor or D-tensor calculations (see next section) are carried out in the same job, ORCA
automatically prints the orientation between the hyperfine/quadrupole couplings and the molecular g- or
D-tensor. For more information on this see section 9.32.11.
8.13.9 The EPR g-Tensor and the Zero-Field Splitting Tensor
The EPR g-tensor is a property that can be well calculated at the SCF level with ORCA through solution of
the coupled-perturbed SCF equations. Consider the following multi-job input that computes the g-tensor at
three different levels of theory:
! UHF HF Def2-SVP TightSCF SOMF(1X)

%eprnmr gtensor 1 ori -3
end
* int 1 2
O 0 0 0 0 0 0
H 1 0 0 1.1056 0 0
H 1 2 0 1.1056 109.62 0
*
$new_job
! UKS LSD Def2-SVP RI SmallPrint PModel SOMF(1X)
end
* int 1 2
O 0 0 0 0 0 0
H 1 0 0 1.1056 0 0
H 1 2 0 1.1056 109.62 0
*
$new_job
! UKS BP Def2-SVP RI SmallPrint PModel SOMF(1X)
end
* int 1 2
O 0 0 0 0 0 0
H 1 0 0 1.1056 0 0
H 1 2 0 1.1056 109.62 0
*
The statement ori -3 means that the origin for this gauge dependent property is to be taken at the center
of electronic charge. The SOMF(1X) defines the details of the spin-orbit-coupling operator and are explained
in section 9.28.2 (in the present case an accurate variant is specified) Other choices and additional variables
exist and are explained in detail in section 9.28.2.
The output looks like the following. It contains information on the contributions to the g-tensor (relativis-
tic mass correction, diamagnetic spin-orbit term (= gauge-correction), paramagnetic spin-orbit term (=
OZ/SOC)), the isotropic g-value and the orientation of the total tensor.
-------------------
ELECTRONIC G-MATRIX
-------------------
Diagonalized gT*g matrix (sqrt of eigenvalues taken):

2.0101862 -0.0031528 0.0000000
-0.0031529 2.0079392 0.0000000
0.0000000 0.0000000 2.0021428
gel 2.0023193 2.0023193 2.0023193

gRMC -0.0003182 -0.0003182 -0.0003182
gDSO(tot) 0.0000725 0.0001471 0.0001485
gPSO(tot) 0.0000692 0.0035675 0.0102602
---------- ---------- ----------
g(tot) 2.0021428 2.0057157 2.0124097 iso= 2.0067561
Delta-g -0.0001765 0.0033964 0.0100905 iso= 0.0044368
Orientation:
X -0.0000000 -0.5763356 0.8172131
Y -0.0000000 -0.8172131 -0.5763356
Z 1.0000000 -0.0000000 -0.0000000
If the total spin of the system is S >1/2 then the zero-field-splitting tensor can also be calculated and printed.
For example consider the following job on a hypothetical Mn(III)-complex.
! UKS BP86 def2-SVP def2/J SOMF(1X)
%eprnmr DTensor ssandso

DSOC cp # qro, pk, cvw
DSS uno # direct
end
* int 1 5
Mn 0 0 0 0 0 0
O 1 0 0 2.05 0 0
O 1 2 0 2.05 90 0
O 1 2 3 2.05 90 180
O 1 2 3 2.05 180 0
F 1 2 3 1.90 90 90
F 1 2 3 1.90 90 270
H 2 1 6 1.00 127 0
H 2 1 6 1.00 127 180
H 3 1 6 1.00 127 0
H 3 1 6 1.00 127 180
H 4 1 6 1.00 127 0
H 4 1 6 1.00 127 180
H 5 1 6 1.00 127 0
H 5 1 6 1.00 127 180

*
The output documents the individual contributions to the D-tensor which also contains (unlike the g-tensor)
contributions from spin-flip terms.
Some explanation must be provided:
The present implementation in ORCA is valid for HF, DFT and hybrid DFT.
There are four different variants of the SOC-contribution, which shows that this is a difficult property.
We will briefly discuss the various choices.
The QRO method is fully documented [161] and is based on a theory developed earlier. [162] The QRO
method is reasonable but somewhat simplistic and is superseded by the CP method described below.
The Pederson-Khanna model was brought forward in 1999 from qualitative reasoning. [163] It also
contains incorrect prefactors for the spin-flip terms. We have nevertheless implemented the method
for comparison. In the original form it is only valid for local functionals. In ORCA it is extended to
hybrid functionals and HF.
The coupled-perturbed method is a generalization of the DFT method for ZFSs; it uses revised prefactors
for the spin-flip terms and solves a set of coupled-perturbed equations for the SOC perturbation.
Therefore it is valid for hybrid functionals. It has been described in detail. [164]
The DSS part is an expectation value that involves the spin density of the system. In detailed calibration
work [165] it was found that the spin-unrestricted DFT methods behave somewhat erratically and that
much more accurate values were obtained from open-shell spin-restricted DFT. Therefore the UNO
option allows the calculation of the SS term with a restricted spin density obtained from the singly
occupied unrestricted natural orbitals.
The DSS part contains an erratic self-interaction term for UKS/UHF wavefunction and canonical
orbitals. Thus, UNO is recommended for these types of calculations. [166] If the option DIRECT is used
nevertheless, ORCA will print a warning in the respective part of the output.
8.13.10 M
ossbauer Parameters
57
Fe Mossbauer spectroscopy probes the transitions of the nucleus between the I = 12 ground state and the
I = 32 excited state at 14.4 keV above the ground state. The important features of the Mossbauer spectrum
are the isomer shift () and the quadrupole splitting (EQ ). An idealized spectrum is shown in Figure
8.29.
The isomer shift measures the shift in the energy of the -ray absorption relative to a standard, usually Fe
foil. The isomer shift is sensitive to the electron density at the nucleus, and indirectly probes changes in the
bonding of the valence orbitals due to variations in covalency and 3d shielding. Thus, it can be used to probe
oxidation and spin states, and the coordination environment of the iron.
Figure 8.29: An idealized Mossbauer spectrum showing both the isomer shift, , and the quadrupole
splitting, EQ .
The quadrupole splitting arises from the interaction of the nuclear quadrupole moment of the excited state
with the electric field gradient at the nucleus. The former is related to the non-spherical charge distribution
in the excited state. As such it is extremely sensitive to the coordination environment and the geometry of
the complex.
Both the isomer shift and quadrupole splitting can be successfully predicted using DFT methods. The isomer
shift is directly related to the s electron density at the nucleus and can be calculated using the formula
= (0 C) + (8.11)
where is a constant that depends on the change in the distribution of the nuclear charge upon absorption,
and 0 is the electron density at the nucleus [167]. The constants and are usually determined via linear
regression analysis of the experimental isomer shifts versus the theoretically calculated electron density for a
series of iron compounds with various oxidation and spin states. Since the electron density depends on the
functional and basis set employed, fitting must be carried out for each combination used. A compilation of
calibration constants (, and C) for various methods was assembled. [168] Usually an accuracy of better
than 0.10 mm s1 can be achieved for DFT with reasonably sized basis sets.
The quadrupole splitting is proportional to the largest component of the electric field gradient (EFG) tensor
at the iron nucleus and can be calculated using the formula:
1
2 2

1
EQ = eQVzz 1 + (8.12)
2 3
where e is the electrical charge of an electron and Q is the nuclear quadrupole moment of Fe (approximately
0.16 barns). Vxx , Vyy and Vzz are the electric field gradient tensors and , defined as

Vxx Vyy
=
(8.13)
Vzz
is the asymmetry parameter in a coordinate system chosen such that |Vzz | |Vyy | |Vxx |.
An example of how to calculate the electron density and quadrupole splitting of an iron center is as follows:
%eprnmr
nuclei = all Fe {fgrad, rho}
end
If the core properties basis set CP(PPP) is employed, one has to increase the radial integration accuracy for
the iron atom, for example:
%basis newgto Fe "CP(PPP)" end

end
%method
SpecialGridAtoms 26
SpecialGridIntAcc 7
end
The output file should contain the following lines, where you obtain the calculated quadrupole splitting
directly and the RHO(0) value (the electron density at the iron nucleus). To obtain the isomer shift one has
to insert the RHO(0) value into the appropriate linear equation (Eq. 8.11).
Moessbauer quadrupole splitting parameter (proper coordinate system)

e**2qQ = -0.406 MHz = -0.035 mm/s
eta = 0.871
Delta-EQ=(1/2e**2qQ*sqrt(1+1/3*eta**2) = -0.227 MHz = -0.020 mm/s
RHO(0)= 11581.352209571 a.u.**-3 # the electron density at the Fe nucleus.
NOTE:
Following the same procedure, Mossbauer parameters can be computed with the CASSCF wavefunction.
In case of a state-averaged CASSCF calculation, the averaged density is used in the subsequent
Mossbauer calculation.
8.13.11 Broken-Symmetry Wavefunctions and Exchange Couplings
A popular way to estimate the phenomenological parameter JAB that enters the HeisenbergDiracvan
Vleck Hamiltonian which parameterizes the interaction between two spin systems is the broken-symmetry
formalism. The phenomenological Hamiltonian is:
~A S
HHDvV = 2JAB S ~B (8.14)
It is easy to show that such a Hamiltonian leads to a ladder of spin states from S = SA + SB down to
S = |SA SB |. If the parameter JAB is positive the systems A and B are said to be ferromagnetically
coupled because the highest spin-state is lowest in energy while in the case that JAB is negative the coupling
is antiferromagnetic and the lowest spin state is lowest in energy.
In the broken symmetry formalism one tries to obtain a wavefunction that breaks spatial (and spin) symmetry.
It may be thought of as a poor mans MC-SCF that simulates a multideterminantal character within a
single determinant framework. Much could be said about the theoretical advantages, disadvantages, problems
and assumptions that underly this approach. Here, we only want to show how this formalism is applied
within ORCA.
For NA unpaired electrons localized on site A and NB unpaired electrons localized on a site B one can
calculate the parameter JAB from two separate spin-unrestricted SCF calculations: (a) the calculation for
the high-spin state with S = (NA +N2
B)
and (b) the broken symmetry calculation with MS = (NA N 2
B)
that features NA spin-up electrons that are quasi-localized on site A and NB spin-down electrons that are
quasi-localized on site B. Several formalisms exist to extract JAB : [169174].
(EHS EBS )
JAB = 2 (8.15)
(SA + SB )
(EHS EBS )
JAB = (8.16)
(SA + SB ) (SA + SB + 1)
(EHS EBS )
JAB = (8.17)
hS 2 iHS hS 2 iBS
We prefer the last definition (Eq. 8.17) because it is approximately valid over the whole coupling strength
regime while the first equation implies the weak coupling limit and the second the strong coupling limit.
In order to apply the broken symmetry formalism use:
%scf BrokenSym NA,NB

end
The program will then go through a number of steps. Essentially it computes an energy and wavefunction for
the high-spin state, localizes the orbitals and reconverges to the broken symmetry state.
CAUTION: Make sure that in your input coordinates site A is the site that contains the
larger number of unpaired electrons!
Most often the formalism is applied to spin coupling in transition metal complexes or metal-radical coupling
or to the calculation of the potential energy surfaces in the case of homolytic bond cleavage. In general,
hybrid DFT methods appear to give reasonable semiquantitative results for the experimentally observed
splittings.
As an example consider the following simple calculation of the singlettriplet splitting in a stretched Li2
molecule:
#
# Example of a broken symmetry calculation
#
! UKS B3LYP/G SVP TightSCF Grid4 NoFinalGrid
%scf BrokenSym 1,1
end
* xyz 0 3
Li 0 0 0
Li 0 0 4
*
There is a second mechanism for generating broken-symmetry solutions in ORCA. This mechanism uses
the individual spin densities and is invoked with the keywords FlipSpin and FinalMs. The strategy is to
exchange the and spin blocks of the density on certain user-defined centers after converging the high-spin
wavefunction. With this method arbitrary spin topologies should be accessible. The use is simple:
#
# Example of a broken symmetry calculation using the "FlipSpin" feature
#
%scf
FlipSpin 1
# Flip spin is a vector and you can give a list of atoms
# on which you want to have the spin flipped. For example
# FlipSpin 17,38,56
# REMEMBER: counting starts at atom 0!
FinalMs 0
# The desired Ms value of the broken symmetry determinant.
# This value MUST be given since the program cannot determine it by itself.
end
* xyz 0 3
Li 0 0 0
Li 0 0 4
*
Finally, you may attempt to break the symmetry by using the SCF stability analysis feature (see Section 9.7).
The ground spin state can be obtained by diagonalizing the above spin Hamiltonian through ORCA-ECA
utility (see 9.32.13).
#
# Example of a broken symmetry calculation using the "FlipSpin" feature
#

%scf
FlipSpin 1
# Flip spin is a vector and you can give a list of atoms
# on which you want to have the spin flipped. For example
# FlipSpin 17,38,56
# REMEMBER: counting starts at atom 0!
FinalMs 0
# The desired Ms value of the broken symmetry determinant.
# This value MUST be given since the program cannot determine it by itself.
end
* xyz 0 3
Li 0 0 0
Li 0 0 4
*
Finally, you may attempt to break the symmetry by using the SCF stability analysis feature (see Section 9.7).
The ground spin state can be obtained by diagonalizing the above spin Hamiltonian through ORCA-ECA
utility (see 9.32.13).
8.13.12 Natural Orbitals for Chemical Valence
In ORCA chemical bonds can be analyzed in terms of the electron density rearrangement taking place upon
bond formation. This quantity is defined as the difference between the electron density of an adduct AB
and that of the so-called supermolecule, obtained from the antisymmetrized product of the wavefunctions
of the non-interacting A and B fragments. This electron density difference can be analyzed by exploiting
the properties of the Natural Orbitals for Chemical Valence [175] (NOCVs), that are the eigenvectors of the
corresponding one electron density difference matrix. This analysis is available in ORCA for HF and DFT
(only for closed shell systems).
The generation of NOCVs is done in three steps.
In the first step, one has to perform single point energy calculations on the isolated fragments (A and B in
the present example) and save the corresponding .gbw files (A.gbw and B.gbw).
In the second step, one has to generate the .gbw file of the supermolecule (AB.gbw), using:
orca_mergefrag A.bgw B.gbw AB.gbw
In the third step, one has to run an SCF calculation using the orbitals in the AB.gbw file as starting guess
and setting the EDA keyword to true in the SCF block.
! blyp def2-tzvpp verytightscf grid5 moread

%moinp "AB.gbw"
%scf
EDA true
end
*xyz 0 1
---AB coordinates---
*
Note that, when the NOCV analysis is used in conjuction with DFT calculations, a very fine integration grid
must be used. With the grid5 keyword, one obtains results that are typically converged with respect to the
intergration grid for standard DFT. For scalar relativistic ZORA calculations, it is advisable to set also IntAcc
6 in the %method block. If the NOCV analysis is requested but a starting guess in not given, the analysis will
be performed with the default guess used in the SCF calculation, leading to completely different results.
A typical output of the analysis is reported the following:
-----------------------------------------------------------
ENERGY DECOMPOSITION ANALYSIS OF THE SCF INTERACTION ENERGY
-----------------------------------------------------------
Delta Total Energy (Kcal/mol) : -9.998
-----------------------------------------------------------
NOCV/ETS analysis
-----------------------------------------------------------
positive eigen. (e) negative eigen.(e) DE_k (Kcal/mol)
-0.1739401311 0.1739401311 -5.952

-0.0607716293 0.0607716293 -1.180
-0.0420733869 0.0420733869 -1.278
-0.0288977718 0.0288977718 -0.490
-0.0207403772 0.0207403772 -0.220
-0.0192167553 0.0192167553 -0.219
-0.0126925147 0.0126925147 -0.148
-0.0077091386 0.0077091386 -0.043
-0.0064403861 0.0064403861 -0.329
-0.0042036490 0.0042036490 -0.139
-0.0000000000 0.0000000000 -0.000
Consistency Check Sum_k DE_k : -9.998
NOCV were saved in : jobname.nocv.gbw
The NOCV eigenvalues are printed alongside with the corresponding energy contributions, computed using the
Extended Transition State (ETS) method of Ziegler [175]. The NOCVs, with the corresponding eigenvalues,
are stored in a .gbw file and can be used for subsequent analyses. Delta Total Energy is the energy
difference between the reference (the supermolecule) and the converged SCF wavefunction. This quantity is
also called Orbital Interaction term in the NOCV/ETS energy decomposition scheme. In order to check if
results are at convergence with respect to the integration grid, one can compare this value with the difference
between the energy of the converged SCF wavefunction and the first step of the SCF iteration. These two
numbers should be identical.
8.13.13 Local Energy Decomposition
The Local Energy Decomposition (LED) analysis [176, 177] is a tool for obtaining insights into the nature
of intermolecular interactions by decomposing the DLPNO-CCSD(T) energy into physical meaningful contri-
butions. By defining two fragments, the corresponding interaction energy is decomposed (see Section 9.28.4
for further information).
All that is required to obtain this decomposition in ORCA is to define the fragments and specify the !LED
keyword in the simple input line. Note that the decomposition of the Hartree-Fock energy makes use of the
RI approximation. Hence, in order to obtain fully consistent results, the initial Hartree-Fock calculation
should also be done with the RI-JK approximation. Otherwise, there will be a slight mismatch between the
SCF energy and the energy decomposition. This may be perfectly acceptable. Note that tightly converged
SCF and coupled cluster energies are preferable in order to avoid any numerical noise in the results. Moreover,
for weakly interacting systems, TightPNO settings should be used. An example of input follows.
! DLPNO-CCSD(T) cc-pVDZ cc-pVDZ/C cc-pVDZ/JK RIJK TightPNO LED
* int 0 1
C(1) 0 0 0 0.000000 0.000 0.000
H(1) 1 0 0 1.095136 0.000 0.000
H(1) 1 2 0 1.094874 109.545 0.000
H(1) 1 2 3 1.094834 109.598 120.042
H(1) 1 2 3 1.095358 109.416 240.067
H(2) 4 1 2 3.110450 138.295 175.624
C(2) 6 4 1 1.095116 84.968 7.763
H(2) 7 6 4 1.095143 109.379 69.448
H(2) 7 6 4 1.095288 109.366 189.600
H(2) 7 6 4 1.094884 109.475 309.523
*
The program will then print information on the fragments and the assignment of localized MOs to fragments.
-------------------------------------
FRAGMENT ANALYSIS OF THE TOTAL ENERGY
-------------------------------------
Number of fragments = 2
Fragment 1: 0 1 2 3 4
Fragment 2: 5 6 7 8 9
Populations of internal orbitals onto Fragments:

0: 1.000 0.000 assigned to fragment 1
This is followed by the decomposition of the Hatree-Fock energy into intra- and inter-fragment contributions,
based on the localisation of the occupied orbitals.
---------------------------------
HARTREE-FOCK ENERGY DECOMPOSITION
---------------------------------
Nuclear repulsion = 40.419238613944

One electron energy = -186.512111063119 (T= 80.154624954772, V= -239.312296909896)
Two electron energy = 28.456717893997 (J= 34.437486290425, K= -5.980768396428)
----------------------
Total energy = -117.636154555178
Consistency check = -117.636154555178 (sum of intra- and inter-fragment energies)
Kinetic energy = 80.154624954772

Potential energy = -197.790779509950
----------------------
Virial ratio = 2.467615307558
------------------------------------------
INTRA-FRAGMENT SCF ENERGY FOR FRAGMENT 1
------------------------------------------

----------------------
Total energy = -44.006594757862

----------------------
Virial ratio = 2.098038719427
------------------------------------------
INTRA-FRAGMENT SCF ENERGY FOR FRAGMENT 2
------------------------------------------


----------------------
Total energy = -65.746253663035

----------------------
Virial ratio = 2.640491328117
------------------------------------------
FRAGMENT SCF INTERACTION FRAGMENTs 2/ 1
------------------------------------------

Nuclear attraction = -27.354439107995
Coulomb repulsion = 5.798302134568
----------------------
Sum of electrostatics = -7.883050043723 (-4946.678 kcal/mol)
Two electron exchange = -0.000256090558 ( -0.161 kcal/mol)

----------------------
Total SCF interaction = -7.883306134280 (-4946.839 kcal/mol)
Sum of INTRA-fragment SCF energies = -109.752848420898

Sum of INTER-fragment SCF energies = -7.883306134280
---------------------
Total SCF energy = -117.636154555178
In order to convert the intra-fragment energy components into contributions to the binding energy, single
point energy calculations must be carried out on the isolated monomers, frozen in the geometry they have in
the adduct, and the corresponding terms must be subtracted. Note that one can also include the deformation
energy by simply computing the energy of the geometrically relaxed fragments (see Section 9.28.4 for further
information). This can be readily achieved with a shell script or a spreadsheet program.
In the next part of the output, a first decomposition of the correlation energy intro intra- and inter-fragment
contributions, based on the localisation of the occupied orbitals, is printed.
--------------------------------
CORRELATION ENERGY DECOMPOSITION
--------------------------------
--------------------------------------------------
INTER- vs INTRA-FRAGMENT CORRELATION ENERGIES (Eh)
--------------------------------------------------
Fragment 1 Fragment 2
---------------------- ----------------------
CCSD sum of pair en. -0.185125133060 -0.185130452302 sum= -0.370255585362
Triples corrections -0.003671899888 -0.003670871693 sum= -0.007342771581
PNO & local corrections -0.000004349165 -0.000004270539 sum= -0.000008619704
---------------------- ----------------------
-0.188801382113 -0.188805594534 sum= -0.377606976647
Interaction correlation for Fragments 2 and 1:

--------------------------------------------------
CCSD pair energies -0.000830541670 ( -0.521 kcal/mol)
Triples contribution -0.000085793870 ( -0.054 kcal/mol)
PNO & Local corrections -0.000013565074 ( -0.009 kcal/mol)
----------------------
Total interaction -0.000929900614 ( -0.584 kcal/mol)
Sum of INTRA-fragment correlation energies = -0.377606976647

Sum of INTER-fragment correlation energies = -0.000929900614
---------------------
Total correlation energy = -0.378536877261
For the DLPNO-CCSD strong pairs, a more sophisticated decomposition, based on the localisation of both
occupied orbitals and PNOs, is also carried out and printed. Accordingly, the correlation energy is divided into
intra-fragment, dispersion and charge transfer components. Note that, due to the charge transfer excitations,
these intra-fragment contributions differ from the ones obtained in the previous decomposition. For further
information, see Section 9.28.4.
--------------------------------------------
Decomposition of the CCSD pair energies
--------------------------------------------
Pipek-Mezey localization is used for localizing PNOs
Decomposition of the intra-fragment CCSD energies ..
FRAGMENT 1 = -0.185125133060 Eh ... done

FRAGMENT 2 = -0.185130452302 Eh ... done
Decomposition of the intra-fragment CCSD energies ..
FRAGMENTS 2,1 = -0.000830541670 Eh ... done
Summary strong pair decomposition

------------------------------------
Intra fragment contributions:
INTRA Fragment 1 -0.185040444
INTRA Fragment 2 -0.185033776
Dispersion contributions:
Dispersion 2,1 -0.000723383
Charge transfer contributions:

Charge Transfer 1 to 2 -0.000139278
Charge Transfer 2 to 1 -0.000149245
8.14 QM/MM Calculations
ORCA is easy to interface as a QM engine in pretty much any QM/MM environment, as it will accept a set
of point charges from an external file (see section 9.1.3) and it will return, in a transparent format, all the
required information for computing energies and gradients to the driving program. In our research group we
have experience with four different QM/MM environments: ChemShell, Gromacs, pDynamo and NAMD. In
the following each of the interfaces are described. Is beyond the scope of the manual to be exhaustive, and
only minimal working examples are going to be presented. These will cover mainly the technical aspects with
respect to the QM part of the QM/MM calculation. For the initial preparation of the system the user is
referred to the documentation of the driving program.
8.14.1 ORCA and Gromacs
In cooperation with the developers of Gromacs software package we developed an interface to ORCA. The
interface is available starting with Gromacs version 4.0.4 up to version 4.5.5.
As mentioned above, the initial setup has to be done with the Gromacs. In the QM/MM calculation Gromacs
will call ORCA to get the energy and the gradients of the QM atoms. The interface can be used to perform
all job types allowed by the Gromacs software package, e.g. optimizations, molecular dynamics, free energy.
In addition, for geometry optimizations we have implemented a microiterative scheme that can be requested
by setting the keyword bOpt = yes in the Gromacs .mdp file. This will cause ORCA to perform a QM
geometry optimization every time it is called by Gromacs. During this optimization ORCA will also compute
the Lennard-Jones interaction between the QM and MM atoms, and freeze the boundary atoms. This
microiterative scheme can also be used to perform a transition state optimization. If you are looking for a
transition state and have a good initial guess for the structure, you can carry out an optimization of the MM
system and at the same time perform a transition state optimization of the QM system with ORCA via the
microiterative scheme. Since it is expensive to calculate the Hessian for each microiterative microiterative
step, the user can tell ORCA to use the (updated) Hessian matrix of the previous step via read temp Hess
in the ORCA input.
In order to allow full flexibility to the user, the information for the QM run are provided in a separate plain
text file, called GromacsBasename.ORCAINFO. When Gromacs writes the input for the ORCA calculation, it
will merge this file with the information on the coordinates, point charges, Lennard-Jones coefficients and
type of calculation (EnGrad, Opt, TSOpt).
Below is a typical example of an input file created by Gromacs, where for each Gromacs optimization step, a
full optimization of the QM-part will be performed by ORCA, instead of just doing the energy and gradient
calculation.
# Optimization step in the Lennard-Jones and point charges field of the MM atoms
! QMMMOpt
# file containing the Lennard Jones coefficients for the Lennard-Jones interaction
%LJCoefficients "temp.LJ"
# file containing the point charges for the electrostatic interaction
8.14 QM/MM Calculations 271
%pointcharges "temp.pc"
%geom
# calculate the exact Hessian before the first optimization step
Calc_Hess true
# in case of a TS optimization the updated Hessian of the previous
# TS optimization run is read instead of calculating a new one
read_temp_Hess true
end
NOTES:
If you are using bOpt or bTS you have to take care of additional terms over the boundary. Either you
can zero out the dihedral terms of the Q2-Q1-M1-M2 configurations, or you can fix the respective Q2
atoms.
If you want to use the ORCA constraints, you have to also put them in the Gromacs part of the
calculation.
If there are no bonds between the QM and MM systems, the bOpt optimization of the QM system
might have convergence problems. This is the case if the step size is too large and thus QM atoms come
too close to MM atoms. The convergence problems can be circumvented by adding extra coordinates
and Hessian diagonal values for Cartesian coordinates of selected QM atoms to the set of internal
coordinates. This should be done for only few atoms that are in the boundary region.
%geom
# add the Cartesian position of atoms 2 and 4 to the
# set of internal coordinates with a diagonal Hessian value of 0.1
Hess_Internal
{C 2 D 0.1}
{C 4 D 0.1}
end
end
8.14.2 ORCA and pDynamo
The interface with the pDynamo library is briefly documented here. It uses the same plain text files to
exchange information between the pDynamo library and ORCA. In order to have simiar functionality as
presented above, we have extended the interface to generate more complex ORCA input files by accepting
verbatim blocks of text. We have also implemented in pDynamo the capability of writing files containing
Lennard-Jones parameters.
A simple example which calculates the QM/MM energy for a small designed peptide in which the side chain
of one amino acid is treated QM is ilustrated below. For this example, a complete geometry optimization is
going to be performed during the ORCA call.
import os
import pCore
import pBabel
import pMolecule
import pMoleculeScripts
# Read the initial coordinates from the .pdb file.

system = pMoleculeScripts . PDBFile_ToSystem (
"1UAO.pdb", modelNumber =1, useComponentLibrary =True)
# Instantiate the required models .

mmmodel = pMolecule . MMModelOPLS (" protein ")
nbmodel = pMolecule . NBModelORCA ()
qcmodel = pMolecule . QCModelORCA (
command =os. getenv (" ORCA_COMMAND "),
deleteJobFiles =False , header ="bp86 def2 -svp qmmmopt / pdynamo ",
job=" chignolin ", run=True)
# Assign the models to the system .

system . DefineMMModel ( mmmodel )
system . DefineQCModel (
qcmodel , qcSelection =pCore. Selection ([35 , 36, 37, 34, 40, 41]))
system . DefineNBModel ( nbmodel )
system . electronicState = pMolecule . ElectronicState (
charge =-1, multiplicity =1)
# Print a summary and calculate the energy .

system . Summary ()
system . Energy ()
After the execution of the above Python program, a series of files are going to be created chignolin.inp,
chignolin.pc, chignolin.lj and ORCA is going to be called. The generated ORCA input file is listed
below.
! bp86 def2-svp qmmmopt/pdynamo

% geom
constraints
{C 0 C}
{C 1 C}
end
end
% pointcharges "chignolin.pc"
% ljcoefficients "chignolin.lj"
* xyz -1 1
H -1.0637532468 1.1350324675 2.4244220779
C -0.5230000000 0.6870000000 3.2490000000
C 0.4180000000 1.7240000000 3.8660000000
O -0.0690000000 2.7620000000 4.2830000000

O 1.6090000000 1.4630000000 3.9110000000
H -1.2240000000 0.3460000000 3.9970000000
H 0.0550000000 -0.1510000000 2.8890000000
*
There are few points that have to be raised here. Because the keyword qmmm/pdynamo was specified in the
header variable, the pDynamo library will automatically add the constraint block in the ORCA input,
which will freeze the link atoms and the QM atoms to which they are bound. It will also generate the
chignolin.lj file containing all the Lennard-Jones parameters. The important parts of this file, which is
somewhat different than the one generated by Gromacs, are listed next.
# number of atoms combination rule

138 0
# x y z sigma epsilon id
-6.778000 -1.424000 4.200000 3.250000 0.711280 -1
-6.878000 -0.708000 2.896000 3.500000 0.276144 -1
-5.557000 -0.840000 2.138000 3.750000 0.439320 -1
...
0.433000 0.826000 0.502000 2.960000 0.878640 -1
-0.523000 0.687000 3.249000 3.500000 0.276144 1
0.418000 1.724000 3.866000 3.750000 0.439320 2
-0.069000 2.762000 4.283000 2.960000 0.878640 3
1.609000 1.463000 3.911000 2.960000 0.878640 4
-2.259000 -0.588000 1.846000 0.000000 0.000000 -1
-1.795000 2.207000 2.427000 2.500000 0.125520 -1
-1.224000 0.346000 3.997000 2.500000 0.125520 5
0.055000 -0.151000 2.889000 2.500000 0.125520 6
-0.311000 2.922000 0.557000 3.250000 0.711280 -1
...
-1.387000 -2.946000 5.106000 2.500000 0.125520 -1
# number of special pairs
22
# atom1 atom2 factor
34 32 0.000000
35 39 0.500000
40 31 0.000000
41 30 0.500000
41 32 0.500000
36 31 0.500000
40 32 0.500000
40 39 0.500000
34 31 0.000000
35 30 0.500000
34 11 0.500000
34 38 0.500000
41 31 0.000000
37 31 0.500000
34 33 0.500000
34 39 0.000000
40 30 0.500000
41 39 0.500000
34 30 0.000000
35 31 0.000000
34 42 0.500000
35 32 0.500000
The second number on the first line refers to the type of combination rule used to calculate the Lennard-Jones
interaction. It is 0 if a geometric average is used (OPLS force field), or 1 for the Lorentz-Berthelot rules
(AMBER force field). The id on the last column is -1 for MM atoms and is equal to the atom number for
the QM atoms. In this case the hydrogen link atom is atom 0. The last block of the file is composed of atom
pairs and a special factor by which their Lennard-Jones interaction is scaled. In general this factor is equal to
1, but for atoms one or two bonds apart is zero, while for atoms three bonds apart depends on the type of
force field, and in this case is 0.5.
After successful completion of the ORCA optimization run, the information will be relayed back the pDynamo
library, which will report the total QM/MM energy of the system. At this point the type QM/MM of
calculation is limited only by the capabilities of the pDynamo library, which are quite extensive.
8.14.3 ORCA and NAMD
Since version 2.12, NAMD is able to perform hybrid QM/MM calculations. A more detailed explanation of
all available key words, setting up the calculation and information on tutorials and on the upcoming graphic
interface to VMD are available on the NAMD website.
Similar to other calculations with NAMD, the QM/MM is using a pdb file to control the active regions. An
example is shown below, where the sidechain of a histidine protonated at N is chosen to be the QM region.
Either the occupancy column or the b-factor column of the file are used to indicate which atom are included
in a QM area and which are treated by the forcefield. In the other column, atoms which are connecting the
QM area and the MM part are indicated similarly. To clarify which column is used for which purpose, the
keywords qmColumn and qmBondColumn have to be defined in the NAMD input.
...
ATOM 1737 CA HSE P 117 14.762 47.946 31.597 1.00 0.00 PROT C
ATOM 1738 HA HSE P 117 14.751 47.579 32.616 0.00 0.00 PROT H
ATOM 1739 CB HSE P 117 14.129 49.300 31.501 1.00 1.00 PROT C
ATOM 1740 HB1 HSE P 117 14.407 49.738 30.518 0.00 1.00 PROT H
ATOM 1741 HB2 HSE P 117 13.024 49.194 31.509 0.00 1.00 PROT H
ATOM 1742 ND1 HSE P 117 13.899 51.381 32.779 0.00 1.00 PROT N
ATOM 1743 CG HSE P 117 14.572 50.261 32.582 0.00 1.00 PROT C
ATOM 1744 CE1 HSE P 117 14.615 52.043 33.669 0.00 1.00 PROT C
ATOM 1745 HE1 HSE P 117 14.356 53.029 34.064 0.00 1.00 PROT H
ATOM 1746 NE2 HSE P 117 15.678 51.318 33.982 0.00 1.00 PROT N
ATOM 1747 HE2 HSE P 117 16.369 51.641 34.627 0.00 1.00 PROT H
ATOM 1748 CD2 HSE P 117 15.706 50.183 33.335 0.00 1.00 PROT C
ATOM 1749 HD2 HSE P 117 16.451 49.401 33.388 0.00 1.00 PROT H
ATOM 1750 C HSE P 117 13.916 47.000 30.775 0.00 0.00 PROT C
ATOM 1751 O HSE P 117 12.965 46.452 31.334 0.00 0.00 PROT O
...
NOTES:
If one wants to include more than one QM region, integers bigger than 1 can be used to define the
different regions.
Charge groups cannot be split when selecting QM and MM region. The reason is that non-integer
partial charges may occur if a charge group is split. Since the QM partial charges are updated in every
QM iteration, this may lead to a change in the total charge of the system over the course of the MD
simulation.
The occupancy and b-factor columns are used for several declarations in NAMD. If two of these come
together in one simulation, the keyword qmParamPDB is used to define which pdb file contains the
information about QM atoms and bonds.
To simplify the selection of QM atoms and writing the pdb file a set of scripts is planned to be included
in future releases of NAMD.
To run the calculation, the keyword qmForces on must be set. To select ORCA qmSoftware "orca" must
be specified and the path to the executables must be given to qmExecPath, as well as a directory where
the calculation is carried out (qmBaseDir). To pass the method and specifications from NAMD to ORCA
qmConfigLine is used. These lines will be copied to the beginning of the input file and can contain both
simple input as well as block input. To ensure the calculation of the gradient, the engrad keyword should be
used.
The geometry of the QM region including the selected links as well as the MM point charges are copied to the
ORCA inputfile automatically. Multiplicity and charge can be defined using qmMult and qmCharge, although
the latter can be determined automatically by NAMD using the MM parameters. It should be noted at this
point that NAMD is capable to handle more than one QM region per QM/MM calculation. Therefore for
each region, charge and multiplicity are expected. In the case of only one QM region, the input looks like the
following:
qmMult "1 1"

qmCharge "1 0"
Currently, two charge modes are available: Mulliken and CHELPG. They have to be specified in the NAMD
input using QMChargeMode and in the qmConfigLine, respectively. Different embedding schemes, point charge
schemes and switching functions are available, which will be not further discussed here. Another useful tool
worth mentioning is the possibility to call secondary executables before the first or after each QM software
execution using QMPrepProc or QMSecProc, respectively. Both are called with the complete path and name
to the QM input file, allowing e.g. storage of values during an QM/MM-MD.
It is strongly enphasized that at this points both programs are constantly developed further. For the latest
information, either the ORCA forum or the NAMD website should be consulted.
277
9 Detailed Documentation
9.1 More on Coordinate Input
We will now enter the detailed discussion of the features of ORCA. Note that some examples are still written
in the old syntax but that there is no need for the user to adopt that old syntax. The new one works as
well.
9.1.1 Fragment Specification
The atoms in the molecule can be assigned to certain fragments. Presently this doesnt do much but to help
to organize the output in the population analysis section (and can therefore be very useful). To assign a
given atom to a given fragment put a (n) directly after the atomic symbol. Fragment enumeration starts
with fragment 1!
%coords
CTyp xyz # the type of coordinates xyz or internal
Charge -2 # the total charge of the molecule
Mult 2 # the multiplicity = 2S+1
coords
Cu(1) 0 0 0
Cl(2) 2.25 0 0
Cl(2) -2.25 0 0
Cl(2) 0 2.25 0
Cl(2) 0 -2.25 0
end
end
In this example the fragment feature is used to divide the molecule into a metal and a ligand fragment
and consequently the program will print the metal and ligand characters contained in each MO in the
population analysis section.
9.1.2 Defining Geometry Parameters and Scanning Potential Energy Surfaces
ORCA lets you define the coordinates of all atoms as functions of user defined geometry parameters. By
giving not only a value but a range of values (or a list of values) to this parameters potential energy surfaces
278 9 Detailed Documentation
can be scanned. In this case the variable RunTyp is automatically changed to Scan. The format for the
parameter specification is straightforward:
%coords
CTyp internal
Charge 0
Mult 1
pardef
rCH = 1.09; # a C-H distance
ACOH = 120.0; # a C-O-H angle
rCO = 1.35, 1.10, 26; # a C-O distance that will be scanned
end
coords
C 0 0 0 0 0 0
O 1 0 0 {rCO} 0 0
H 1 2 0 {rCH} {ACOH} 0
H 1 2 3 {rCH} {ACOH} 180
end
end
In the example above the geometry of formaldehyde is defined in internal coordinates (the geometry functions
work exactly the same way with cartesian coordinates). Each geometrical parameter can be assigned as a
function of the geometry parameters by enclosing the function braces, {}. For example a function may
look like {0.5*cos(Theta)*rML+R}. Note that all trigonometric functions expect their arguments to be in
degrees and not radians. The geometry parameters are expected to be defined such that the lengths come
out in
Angstroms and the angles in degrees. After evaluating the functions the coordinates will be converted
to atomic units. In the example above the variable rCO was defined as a Scan parameter. Its value will
be changed in 26 steps from 1.3 down to 1.1 and at each point a single point calculation will be done. At
the end of the run the program will summarize the total energy at each point. This information can then
be copied into the spreadsheet of a graphics program and the potential energy surface can be plotted. Up
to three parameters can be scan parameters. In this way grids or cubes of energy (or property) values as a
function of geometry can be constructed.
If you want to define a parameter at a series of values rather than evenly spaced intervals, the following
syntax is to be used:
%coords
CTyp internal
Charge 0
Mult 1
pardef
rCH = 1.09; # a C-H distance
ACOH= 120.0; # a C-O-H angle
rCO [1.3 1.25 1.22 1.20 1.18 1.15 1.10]; # a C-O distance that will be scanned
end
9.1 More on Coordinate Input 279
coords
C 0 0 0 0 0 0
O 1 0 0 {rCO} 0 0
H 1 2 0 {rCH} {ACOH} 0
H 1 2 3 {rCH} {ACOH} 180
end
end
In this example the C-O distance is changed in seven non-equidistant steps. This can be used in order
to provide more points close to a minimum or maximum and fewer points at less interesting parts of the
surface.
A special feature has also been implemented into ORCA - the parameters themselves can be made functions
of the other parameters as in the following (nonsense) example:
%coords
CTyp internal
Charge 0
Mult 1
pardef
rCOHalf= 0.6;
rCO = { 2.0*rCOHalf };
end
coords
C 0 0 0 0 0 0
O 1 0 0 {rCO} 0 0
O 1 0 0 {rCO} 180 0
end
end
In this example the parameter rCO is computed from the parameter rCOHalf. In general the geometry is
computed (assuming a Scan calculation) by (a) incrementing the value of the parameter to be scanned, (b)
evaluating the functions that assign values to parameters and (c) evaluating functions that assign values
to geometrical variables. Although it is not mandatory it is good practice to first define the static or
scan-parameters and then define the parameters that are functions of these parameters.
Finally ORCA has some special features that may help to reduce the computational effort for surface scans:
%method
SwitchToSOSCF true
# switches the converger to SOSCF
# after the first point. SOSCF may
# converge better than DIIS if the
# starting orbitals are good.
# default = false
ReducePrint true # reduce printout after the first
# point default=true
# The initial guess can be changed after the first
# point. The default is MORead. The MOs of the pre-
# vious point will in many cases be a very good guess
# for the next point. In some cases however, you may
# want to be more conservative and use a general guess.
ScanGuess OneElec # the one electron matrix
Hueckel # the extended Hueckel guess
PAtom # the PAtom guess
PModel # the PAtom guess
MORead # MOs of the prev. point
end
NOTE:
You can scan along normal modes of a hessian using the NMScan feature as described in section 9.18.8.9.
The surface scan options are also supported in conjunction with TD-DFT/CIS or MR-CI calculations
(see section 9.18.8.7).
9.1.3 Inclusion of Point Charges
In some situations it is desirable to add point charges to the system. In ORCA there are two mechanisms to
add point-charges. If you only want to add a few point charges you can mask them as atoms as in the
following (nonsense) input:
# A water dimer
! BP86 def2-SVP
* xyz 0 1
O 1.4190 0.0000 0.0597
H 1.6119 0.0000 -0.8763
H 0.4450 0.0000 0.0898
Q -0.834 -1.3130 0.0000 -0.0310
Q 0.417 -1.8700 0.7570 0.1651
Q 0.417 -1.8700 -0.7570 0.1651
*
Here the Qs define the atoms as point charges. The next four numbers are the magnitude of the point
charge and its position. The program will then treat the point charges as atoms with no basis functions and
nuclear charges equal to the Q values.
9.1 More on Coordinate Input 281
If you have thousands of point charges to treat, as in a QM/MM calculation, it is more convenient, and
actually necessary, to read the point charges from an external file as in the following example:
# A water dimer
! BP86 def2-SVP
% pointcharges "pointcharges.pc"
* xyz 0 1
O 1.4190 0.0000 0.0597
H 1.6119 0.0000 -0.8763
H 0.4450 0.0000 0.0898
*
The program will now read the file pointcharges.pc that contains the point-charge information and then
call the module orca pc which adds the point charge contribution to the one-electron matrix and the nuclear
repulsion. The file pointcharges.pc is a simple ASCII file in the following format:
3
-0.834 -1.3130 0.0000 -0.0310
0.417 -1.8700 0.7570 0.1651
0.417 -1.8700 -0.7570 0.1651
The first line gives the number of point charges. Each consecutive line gives the magnitude of the point
charge (in atomic units) and its position (in Angstrom units!). However, it should be noted that ORCA
treats point charges from an external file differently than Q atoms. When using an external point charge
file, the interaction between the point charges is not included in the nuclear energy. This behavior originates
from QM/MM, where the interactions among the point charges is done by the MM program. These programs
typically use an external point charge file when generating the ORCA input. To add the interaction of the
point charges to the nuclear energy, the DoEQ keyword is used either in the simple input or the %method block
as shown below.
# A non QM/MM pointcharge calculation

!DoEQ
% pointcharges "pointcharges.pc"
%method
DoEQ true
end
9.2 Choice of Computational Model
9.2.1 Features Common to All Calculations
The computational model is specified in the block %method. The following choices exist:
%method
Method HFGTO # Hartree-Fock with GTOs (synonym HF)
DFGTO # Density Functional with GTOs (syn. DFT)
MP2 # Second order Moeller-Plesset
CNDO # complete neglect of differential overlap
INDO # intermediate neglect of d. o.
NDDO # neglect of diatomic d. o.
end
In the case of Hartree-Fock calculations [59] nothing else is required in this block. Density functional
calculations [178, 179] need slightly more attention.
The RunType (=type of calculation to be performed) is chosen as follows:
%method
RunTyp Energy # single point calc. (default)
Gradient # single point energy and gradient
Opt # Geometry optimization
MD # Molecular dynamics
Scan # scan of geometric parameters
end
You can tell the main program the explicit names and positions of the other modules. In this way you could
in principle also interface your own programs to ORCA as long as they respect the input/output conventions
used in ORCA (which are, however, reasonably complicated).
%method
#*** the name of the SCF program
ProgSCF "MySCFProg.exe"
#*** the name of the GTO integral program
ProgGTOInt "MyGTOIntProg.exe"
#**** the name of the MP2 module
ProgMP2 "MyProgMP2.exe"
#*** the name of the plot program
ProgPlot "MyPlotProgram.exe"
#*** the name of the SCF gradient program
ProgSCFGrad "MySCFGradientProg.exe"
#*** the name of the geometry relaxation program
9.2 Choice of Computational Model 283
ProgGStep "MyProgGStep.exe"
#*** the name of the molecular dynamics program
ProgMD "MyProgMD.exe"
#*** the name of the moment integral program
ProgMom "MyProgMom.exe"
# *** the name of the EPR/NMR module
ProgEPRNMR "MyProgEPRNMR.exe"
#*** the name of the CP-SCF program
ProgCPSCF "MyProgCPSCF.exe"
# *** the name of the CI-singles and TD-DFT module
ProgCIS "MyProgCIS.exe"
# *** the name of the Relativistics module
ProgREL "MyProgREL.exe"
end
For example, if the executables are all located in the same run directory (and the PATH variable contains
this directory!) use:
%method
ProgSCF "orca_scf.exe"
ProgGTOInt "orca_gtoint.exe"
ProgMP2 "orca_mp2.exe"
ProgPlot "orca_plot.exe"
ProgSCFGrad "orca_scfgrad.exe"
ProgGStep "orca_gstep.exe"
ProgMD "orca_md.exe"
ProgMom "orca_mom.exe"
ProgCPSCF "orca_cpscf.exe"
ProgEPRNMR "orca_eprnmr.exe"
ProgCIS "orca_cis.exe"
ProgRel "orca_rel.exe"
ProgMDCI "orca_mdci.exe"
end
9.2.2 Density Functional Calculations
9.2.2.1 Choice of Functional
Basic Choice of Density Functional. If you are doing a DFT calculation [178, 179], the following
choices for local and gradient corrected density functionals are available (see also simple input keywords in
section 6.2.2):
%method
# Choices for Functional. If no reference is given,
# look further below for the references for individual
# exchange and correlation parts
Functional
#***************************************
# Local functionals
#***************************************
HFS # Hartree-Fock Slater (Slater exchange only)
XAlpha # The famous old Slater Xa theory
LSD # Local spin density (VWN-5A form)
VWN5 # Local spin density (VWN-5)
VWN3 # Local spin density (VWN-3)
PWLDA # Local spin density (PW-LDA)
#***************************************
# Pure GGA functionals
#***************************************
BNULL # Becke 88 exchange, no corr.
BVWN # Becke 88 exchange, VWN-5 corr.
BP # Becke 88 X-Perdew 86 correlation
PW91 # Perdew-Wang GGA-II 91 func.
mPWPW # Modified PW with PW correlation
mPWLYP # same with LYP correlation
BLYP # Becke X with LYP correlation
GP # Gill 96 X, Perdew 86 corr.
GLYP # Gill 96 X with LYP correlation
PBE # Perdew-Burke-Ernzerhof
revPBE # Revised PBE (exchange scaling) [180]
RPBE # Revised PBE (functional form of X) [181]
PWP # PW91 exchange + P86 correlation
OLYP # the optimized exchange and LYP
OPBE # the optimized exchange and PBE
XLYP # the Xu/Goddard exchange and LYP
B97-D # Grimmes GGA including D2 dispersion correction
B97-D3 # Grimmes GGA including D3 dispersion correction
PW86PBE # as used for vdw-DF and related [182] [183] [184]
RPW86PBE # revised version of the exchange functional [185]
#***************************************
# Meta GGA functionals
#***************************************
M06L # Truhlars semi-local functional [186]
TPSS # the TPSS functional
revTPSS # revised TPSS [187] [188]
#***************************************
# Hybrid functionals
#***************************************
B1LYP # One parameter Hybrid of BLYP
B3LYP # Three parameter Hybrid of BLYP [189]
B1P # Analogous with Perdew correlation
B3P # Analogous with Perdew correlation
G1LYP # 1 par. analog with Gill 96 X
G3LYP # 3 par. analog with Gill 96 X
G1P # similar with P correlation
G3P # similar with P correlation
PBE0 # 1 parameter version of PBE [190]
PWP1 # 1 parameter version of PWP (analogous to PBE0)
M06 # Truhlars 2006 low-HF hybrid [191]
M06-2X # Truhlars 2006 high-HF hybrid [191]
mPW1PW # 1 parameter version of mPWPW (analogous to PBE0)
mPW1LYP # 2 parameter version of mPWLYP (analogous to PBE0)
PW91_0 # 1 parameter version of PW91 (analogous to PBE0)
O3LYP # 3 parameter version of OLYP [192]
X3LYP # 3 parameter version of XLYP [193]
PW6B95 # Hybrid functional by Truhlar [194]
B97 # Beckes original hybrid
BHANDHLYP # Half-and-half Becke hybrid functional [195]
#***************************************
# Range-Separated Hybrid functionals
#***************************************
wB97 # Head-Gordons fully variable DF [114]
wB97X # Head-Gordons DF with minimal Fock exchange [114]
wB97X-D3 # Chais refit incl. D3 correction [115]
CAM-B3LYP # Handys fit [112]
LC-BLYP # Hiraos original application [113]
#***************************************
# Meta Hybrid functionals
#***************************************
TPSSh # hybrid version of TPSS with 10% HF exchange
TPSS0 # hybrid version of TPSS with 25% HF exchange
#***************************************
# Double-Hybrid functionals (mix in MP2)
#***************************************
B2PLYP # Grimmes 2006 double-hybrid [106]
mPW2PLYP # Schwabe/Grimme improved double-hybrid [107]
B2GP-PLYP # Martins refit of B2PLYP [196]
B2K-PLYP # Martins refit of B2PLYP [196]
B2T-PLYP # Martins refit of B2PLYP [196]
PWPB95 # Recent Grimme double-hybrid [197]
end
Note that Functional is a compound key. It chooses specific values for the variables Exchange, Correlation
and ACM described below. If given as a simple input keyword, in some cases, it will also activate a dispersion
correction. You can explicitly give these variables instead or in addition to Functional. However, make sure
that you specify these variables after you have assigned a value to Functional or the values of Exchange,
Correlation and ACM will be reset to the values chosen by Functional.
Empirical Parameters in Density Functionals. Some of the functionals incorporate empirical parameters
that can be changed to improve agreement with experiment. In ORCA there is some freedom to change
functional parameters. Currently there are several parameters that can be changed (other than the parameters
used in the hybrid functionals). The first of these parameters is of Slaters X method. Theoretically
it has a value of 2/3 and this is used in the HFS and LSD functionals. However, exchange is about 10%
underestimated by this approximation (a very large number!) and a value around 0.70-0.75 seems to be better
for molecules. The second parameter is the parameter for Beckes gradient corrected exchange functional.
Becke has determined the value 0.0042 by fitting the exchange energies for rare gas atoms. There is some
evidence that with smaller basis sets for molecules a slightly smaller value such as 0.0039 gives improved
results. The final parameter is the value occuring in the PBE exchange functional. It has been given the
value 0.804 by Perdew et al. in order to satisfy the Lieb-Oxford bound. Subsequently other workers have
argued that a larger value for this parameter (around 1.2) gives better energetics and this is explored in
the revPBE functional. Note that it also has been shown that while revPBE gives slightly better energetics
it also gives slightly poorer geometries. Within the PBE correlation functional, there is also the C (not
to be confused with the exchange parameter in Beckes exchange functional). Its original value in the
PBE functional is C = 0.066725, but modified variants exist, e.g., the PBEsol functional, or the PBEh-3c
compound method. Furthermore, the parameter in the PBE exchange functional may be modified. In the
2
original formulation it is related to C via = C 3 , but has been changed in the latter variants as well.
%method
XAlpha 0.75 # Slaters alpha parameter (default 2/3)
XBeta 0.0039 # Beckes beta parameter
# (default 0.0042)
XKappa 0.804 # PBE(exchange) kappa parameter (default 0.804)
CBetaPBE 0.066725 # PBE(correlation) beta (default 0.066725)
XMuePBE 0.21952 # PBE(exchange) mue parameter (default 0.21952)
end
Specifying Exchange and Correlation approximations individually. The following variables are
available for specifying the exchange and correlation approximations individually and to construct user
defined hybrid or extended hybrid functionals:
%method
Exchange X_NOX # no exchange
X_SLATER # Slaters local exchange [198] [199]
X_B88 # Becke 88 gradient exchange [200]
X_G96 # Gill 96 gradient exchange [201]
X_PW91 # Perdew-Wang 91 gradient exchange [202] [203]
X_mPW # Adamo-Barone modification of PW [204]
X_PBE # PBE exchange [184]

X_RPBE # RPBE [181]
X_OPTX # Hoe/Cohen/Handys optimized exchange [205]
X_X # Xu/Goddard [193]
X_TPSS # TPSS meta GGA exchange [206]
X_B97D # Grimmes modified exchange for the B97-D GGA [1]
X_B97BECKE # Beckes original exchange for the B97 hybrid [207]
Correlation C_NOC # no correlation
C_VWN5 # Local VWN-V parameters [208]
C_VWN3 # Local VWN-III parameters [208]
C_PWLDA # Local PW 91 [209]
C_P86 # Perdew 86 correlation [210]
C_PW91 # Perdew-Wang 91 correlation [202]
C_PBE # PBE correlation [184]
C_LYP # LYP correlation [211]
C_TPSS # TPSS meta-GGA correlation [206]
C_B97D # Grimmes modified correlation for the B97-D GGA [1]
C_B97BECKE # Beckes original correlation for the B97 hybrid [207]
# for hybrid functionals. Reference, Becke [195]
ACM ACM-A, ACM-B, ACM-C
# ACM-A: fraction of HF-exchange in hybrid DFT
# ACM-B: scaling of GGA part of DFT exchange
# ACM-C: scaling of GGA part of DFT correlation
# "extended" hybrid functional
ScalLDAC 1.0 # scaling of the LDA correlation part
ScalMP2C 0.0 # fraction of MP2 correlation mixed into
# the density functional
end
Hybrid Density Functionals. The hybrid DFs [195, 212] are invoked by choosing a nonzero value for
the variable ACM. (ACM stands for adiabatic connection model). Specifically, these functionals have the
following form:
X X X C C
EXC = aEHF + (1 a) ELSD + bEGGA + ELSD + cEGGA (9.1)
X X
Here, EXC is the total exchange/correlation energy, EHF is the Hartree-Fock exchange, ELSD is the local
X C
(Slater) exchange, EGGA is the gradient correction to the exchange, ELSD is the local, spin-density based
C
part of the correlation energy and EGGA is the gradient correction to the correlation energy. This brings
use to a little bit awkward subject: several hybrid functionals with the same name give different values in
different programs. The reason for this is that they either choose slightly different default values for the
parameters a, b and c and or they differ in the way they treat the local part of the correlation energy. Different
parameterizations exist. The most popular is due to Vosko, Wilk and Nusair (VWN, [208]). However, VWN
in their classic paper give two sets of parameters - one in the main body (parameterization of RPA results;
known as VWN-III) and one in their table 5 of correlation energies (parameterization of the Ceperley/Alder
Monte Carlo results; known as VWN-V). Some programs choose one set, others the other. In addition a
slightly better fit to the uniform electron gas has been produced by Perdew and Wang [209]. The results
from this fit are very similar to what the parameters VWN5 produce (the fit to the Ceperley Alder results)
whereas VWN3 (fit to the RPA results) produces quite different values. To be short - in ORCA almost all
functionals choose PWLDA as the underlying LDA functional. A special situation arises if LYP is the correlation
functional [211]. LYP itself is not a correction to the correlation but includes the full correlation. It is therefore
used in the B3LYP method as:
C C C C

EB3LYP = ELSD + c ELYP ELSD (9.2)
In ORCA VWN5 is chosen for the local correlation part. This choice is consistent with the TurboMole
program [213215] but not with the Gaussian program [216]. However, the user has full control over this
point. You can choose the underlying local part of any correlation functional with the variable LDAOpt:
%method
LDAOpt C_PWLDA
C_VWN5
C_VWN3
end
Specifying C VWN3 for LDAOpt together with Functional=B3LYP should give results very close to the B3LYP
functional as implemented in the Gaussian series of programs1 .
In particular for the popular B3LYP functional the following aliases are defined in order to facilitate
comparisons with other major electronic structure packages:
%method
Functional B3LYP # consistent with TurboMole
B3LYP_TM # consistent with TurboMole
# = Functional= B3LYP
# LDAOpt = C_VWN5;
B3LYP_G # consistent with Gaussian
# = Functional= B3LYP
# LDAOpt = C_VWN3;
end
One Parameter Hybrid Density Functionals. A few words on the one parameter hybrid methods
appears in order. Through the underlying LDA dependence of the three parameter hybrids different programs
give different answers because they differ in the underlying LDA. On the other hand, it has recently been
argued from theoretical reasoning that the optimal mixing ratio for DFT and HF exchange is 0.25 [218].
Furthermore numerical calculations have shown that the results of using this fixed ratio and not scaling
the GGA correlation or exchange are as good as the original three parameter hybrids [219]. I personally
1
There is some evidence that the version used in the Gaussian program gives miniscule better results in molecular
applications then the TurboMole variant but the differences are very small [217]
sympathize with these ideas because they are based on theory and they remove some arbitrariness from the
hybrid procedures. Also the slightly higher HF-exchange (0.25 in favor of 0.20 used in the original three
parameter hybrids) is, I believe, in the right direction. Thus the one parameter hybrids have the simple
form:
X
+ a0 EHF
X X C

EXC = EDFT EDFT + EDFT (9.3)
with a0 = 14 which is the same as putting: a = a0 , b = 1 a0 and c = 1 in the three parameter hybrids and
this is how it is implemented. The one parameter hybrid PBE0 has been advertised as a hybrid functional of
overall well balanced accuracy [190].
Extended double-hybrid functionals. In addition to mixing the HF-exchange into a given DF, Grimme
has proposed to mix in a fraction of the MP2 correlation energy calculated with hybrid DFT orbitals. [106]
Such functionals may be refered to as extended hybrid functionals. Grimmes expression is:
HF DF T DF T MP 2
EXC = aEX + (1 a) EX + (1 c) EC + cEC (9.4)
Such functionals can be user-defined in ORCA as follows:
%method
ScalHFX = a
ScalDFX = 1-a
ScalGGAC = 1-c
ScalLDAC = 1-c
ScalMP2C = c
end
Grimme recommends the B88 exchange functional, the LYP correlation functional and the parameters a=0.53
and c=0.27. This gives the B2PLYP functional which appears to be a fair bit better than B3LYP based on
Grimmes detailed evaluation study.
Presently, this methodology covers single points and analytic gradients (hence all forms of geometry optimiza-
tion, relaxed scans, transition state searches and numerical frequencies are supported). Note that you need to
choose %mp2 density relaxed end in order to get the correct response density which is consistent with first
order properties as analytic derivatives. By default this density is not calculated since its construction adds
significant overhead to the calculation. Therefore you have to specifically request it if you want to look at the
consistent density. You can also choose %mp2 density unrelaxed end which would give you the unrelaxed
(expectation value like) density of the method at considerably less computational cost. However, this is not
recommended since the changes to the relaxed density are considerable in our experience and the unrelaxed
density has a much weaker theoretical status than its relaxed counterpart.
Range-separated hybrid functionals. ORCA supports functionals based on the error function splitting
of the two-electron operator used for exchange as first realized by Hirao and coworkers [111]:
1 1 1
r12 = erfc( r12 ) r12 + erf( r12 ) r12 (9.5)
| {z } | {z }
SR LR
Rx
where erf(x) = 2 0 exp(t2 )dt and erfc(x) = 1 erf(x). Note that the splitting is only applied to exchange;
one-electron parts of the Hamiltonian, the electron-electron Coulomb interaction and the approximation for
the DFT correlation are not affected. Later, Handy and coworkers generalized the ansatz to: [112]
1 1 [ + erf( r12 )] + erf( r12 )

r12 = 1 + 1 (9.6)
r12 r12
| {z } | {z }
SR LR
The splitting has been described in graphical form (according to Handy and coworkers), along the terminology
of ORCA, in Figure 9.1:
100%
ACM-B
80%
DFT exchange
60%
, RangeSepScal
40%
exact exchange
20%
, ACM-A
0%
0
r12
Figure 9.1: Graphical description of the Range-Separation ansatz. The gray area corresponds to
Hartree-Fock exchange. and follow Handys terminology. [112]
The splitting has been used to define the B97 family of functionals where the short-range part (SR) is
described by DFT exchange and the long-range part by exact exchange/Hartree-Fock exchange. The same is
true for CAM-B3LYP and LC-BLYP. It is possible to use a fixed amount of Hartree-Fock exchange (EXX)
and/or a fixed amout of DFT exchange in this ansatz.
Functional Keyword fixed EXX variable part /bohr-1 fixed DFT-X Reference
B97 WB97 100% 0.40 [114]
B97X WB97X 15.7706% 84.2294% 0.30 [114]
B97X-D3 WB97X-D3 19.5728% 80.4272% 0.25 [115]
CAM-B3LYP CAM-B3LYP 19% 46% 0.33 35% [112]
LC-BLYP LC-BLYP 100% 0.33 [113]
The currently available speed-up options are RIJONX and RIJCOSX. Otherwise, integral-direct single-
point calculations, calculations involving the first nuclear gradient (i.e. geometry optimizations), frequency
calculations, TDDFT, TDDFT nuclear gradient, and EPR/NMR calculations are the only supported job
types thus far. In principle, it is possible to change the amount of fixed Hartree-Fock exchange (ACM-A) and
the amount of variable exchange (RangeSepScal) and , though this is not recommended. The amount of
fixed DFT Exchange (ACM-B) can only be changed for CAM-B3LYP and LC-BLYP. In other words, ACM-B
is ignored by the B97 approaches, because no corresponding -independent exchange functional has been
defined.
! RKS CAM-B3LYP SVP RIJONX def2/J
%method
RangeSepEXX True # must be set
RangeSepMu 0.25 # should not be set to 0.0 or below
RangeSepScal 0.7 # should sum to 1 with ACM-A and ACM-B
ACM 0.2, 0.1, 1.0 # ACM-A, ACM-B, ACM-C
end
* xyz 0 1
H 0.0 0.0 0.0
H 0.0 0.0 0.7
*
Note: For information on the ACM formalism, see preceding section called Specifying Exchange and
Correlation approximations individually. While it is technically possible to choose an exchange functional
that has no -dependence, this makes conceptually no sense.
9.2.2.2 Choice of Integration Grid
The next subject of importance for the DFT calculation is the size of the integration grid for numerical
integration. Three-dimensional numerical quadrature is necessary due to the complicated analytical form of
the exchange-correlation potential (and energy). The expressions for these quantities are so complicated that
there is no hope of finding analytical solutions to the required integrals and some numerical approximation,
commonly three-dimensional numerical integration, is necessary. The numerical integration [220225] is a
major step in any DFT calculation and if the RI approximation [223229] is used for the Coulomb part it
usually dominates the total wall clock time.
Accuracy. The good news about the numerical integration is that the effort only increases approximately
linearly with the molecular size such as to make calculations on relatively large molecules not prohibitive
[222, 230232]. The bad news about numerical integration is that it is not trivial to make it highly accurate.
Especially if you choose relatively small grids it is advisable to check that the results are converged (say to at
least 1 mEh) by reconverging the calculation with a larger grid (Grid) and higher radial integral accuracy
(IntAcc). As a rule of thumb the error in the numerically integrated total electron density is on the same
order of magnitude as the error in the numerically integrated XC energy [221]2 . Therefore the program prints
this information each iteration (Num. Int. El. : . . . ).
Rotational Invariance. Through the very fact that a grid is used, the total energy depends on the
molecular orientation which is of course unphysical [233, 234]. Unless the integration grid is large enough
to make the dependence negligible it is therefore not a good idea to calculate closely related molecules in
grossly different orientations. This unpleasant feature arises because ORCA does not yet have a standard
orientation for molecules like other programs. Note that other programs also have that rotational variance
2
Actually, the XC energy and potential are somewhat smoother than the density itself leading to the expectation
that the error in the density is a pessimistic estimate of the error in the XC energy.
problem but they hide it through employment of a standard orientation. Grid free methods to integrate
the XC part are known [235237] and they remove these problems. However, they will probably not be
implemented in the near future into ORCA because they have a less favorable computational scaling than
the numerical integration, requiring a large additional O(N3 ) diagonalization at each iteration.
Choice of Grid. Several default grids are available in ORCA. They are chosen by specifying the variable
Grid.
%method
Grid 0 # product grid
1 # Lebedev 50 (not recommended)
2 # Lebedev 110 points (default for SCF iterations)
3 # Lebedev 194 points (more accurate)
4 # Lebedev 302 points (default for FinalGrid)
5 # Lebedev 434 points (large)
6 # Lebedev 590 points (larger)
7 # Lebedev 770 points (very large)
end
Like Functional, Grid is merely a compound variable that affects several other variables that control the
details of grid design. Any of these variables can be specified after Grid in order to change the default values.
In general the grid consists of a series of radial shells around each atom and an angular grid for each radial
shell.
Specifying Grid parameters Individually. A particular radial integration method3 is determined by

specifying RadialGrid. In my hands, the default, GaussChebyshev [221, 238], is the most efficient and it is
not recommended to change to EulerMcLaurin [222, 230232] integration. The number of radial points is
determined by the parameter IntAcc. From IntAcc the number of radial shells for a given atom is computed
as (taken from Krack and Koster [238] = 10IntAcc ):
nrad (A) = 5 (3 log nA + 8) (9.7)
where nA is the row of the periodic table of element A. The defaults for IntAcc range from 3.7 to 5.5.4 It
is not recommended to go below 3.5. AngularGrid specifies the largest angular grid to be used (see below).
If SimpleGrid is chosen a product integration is performed over the angular variables with a Gauss-Legendre
integration for and equally spaced points for [222, 230232]. This integration is useful because it is open
ended - the number of points can be increased ad infinitum. It is however, difficult to imagine situations
where the largest Lebedev grid is still too small. The Lebedev grids [239245] are generally more efficient
(theoretically by a factor 1.5). For the SimpleGrid the parameter NThetaMax is the number to multiply nrad
with to obtain the maximum number of points for a given atom. The number of points is automatically
3
For the experts - ORCA always uses the mapping M3 and also the atomic size adjustments of Treutler and Ahlrichs
in the radial integration.
4
In general it is usually the angular integration that limits the accuracy. Thus it is a questionable strategy to choose
a small angular grid and then increase IntAcc to a large value.
chosen for a given based on a recommendation by Treutler and Ahlrichs and avoids crowding of points near
the poles [221].
Finally BFCut and WeightCut are thresholds. If a gridpoint has a weight smaller than WeightCut it is simply
neglected (Becke scheme). If the weight scheme is chosen according to Weight AtomXC, the grid point is also
neglected if it gives a contribution of <WeightCut to the promolecular density which is constructed from a
superposition of spherically symmetric atomic densities. BFCut determines when to neglect the contribution
of a given basis function to a given gridpoint or the contribution of a given basis function product to the
total density at a given gridpoint. There was extensive experimentation with these values and they should
probably not be changed. In particular, the results may be quite sensitive to WeightCut (larger values are
not recommended) but should be quite stable with BFCut (smaller values should not change the energy
appreciably but may significantly raise the computation time).
%method
RadialGrid GaussChebyshev # (default)
EulerMcLaurin
AngularGrid Lebedev50 # =1
Lebedev110 # =2
Lebedev194 # =3 (default)
Lebedev302 # =4
Lebedev434 # =5
Lebedev590 # =6
Lebedev770 # =7
SimpleGrid # =0
IntAcc 5.0 # determines no. of radial points
NThetaMax 0.7 # only for AngularGrid=0
GridPruning 0 # no Pruning
1 # Grid pruning algorithm 1
2 # Grid pruning algorithm 2 (no longer there)
3 # Grid pruning algorithm 3 (default)
4 # Adaptive grid pruning (no longer there)
WeightScheme Weight_Becke # default. The Becke weight scheme
Weight_AtomXC # Choose weights from
# superposition of atomic
# exchange densities. Good choice
# for DFT!
HGridReduced true # Reduce grids for H and He by one
# unit (recommended)
BFCut 1e-10 # basis fcn. cut. Is adjusted according to
# convergence tolerances
WeightCut 1e-14 # grid weight cut. default: 1e-14
end
Grid pruning. Like most other programs ORCA prunes the angular grids [222,230232]. It is well known
that the radial shells close to the nucleus need less angular grid points to achieve high accuracy while those
in the bonding region need larger angular grids. Taking this into account one can save large numbers of grid
points (and therefore computation time) by using smaller angular grids close to the nuclei. There was some
experimentation with this feature and the defaults should probably not be changed.
GridPruning=3 Here the radial integration is divided into five domains that defined through the
numbers 1 to 4 [222, 230232]. There are spheres defined by i Rmax , where Rmax is Clementis radius
of the outermost valence orbital [246]. The angular grids are used in order ng -3, ng -2, ng -1, ng , ng -1
where ng is the size of the largest angular grid. For example for AngularGrid=4 the Lebedev grids
with 50, 110, 194, 302, 194 points are used in the domains one to five. In my experience this leads to
significant reduction in the number of points as well as small errors (on the order of some Eh ) relative
to the unpruned grids of the same size.
Default Grids Having specified the individual parameters, the default grids are characterized by the
following settings:
Gridc AngularGrid IntAcc H,Hed Li-Ned Na-Ard K-Krd

0 SimpleGrid 4.34a 30 35 40 45
1 Lebedev50 4.34 30 35 40 45
2 Lebedev110 4.34 30 35 40 45
3 Lebedev194 4.34 30 35 40 45
4 Lebedev302 4.67 30 35 40 45
5 Lebedev434 5.01 35 40 45 50
6 Lebedev590 5.34 40 45 50 55
7 Lebedev770 5.67 45 50 55 60
a NThetaMax=0.4 for Grid 0.

b The choice of IntAcc was such that there is no further improvement in the accuracy with more radial points
c for all grids GridPruning=3, RadialGrid=GaussChebyshev, HGridReduce=true (except grid 0)
d Number of radial points
If no value for Grid is specified in the input, Grid=2 is used for the SCF iterations and Grid=4 for the final
energy (vide infra).
Multigrid feature. The TurboMole developers [213215] have advocated a method where the SCF iterations
are done with a small grid and gradients and final energies are evaluated on a larger, more accurate grid [221].
Overall this gives a substantial improvement in speed with no significant loss in accuracy as long as the
smaller grid for the SCF iterations is large enough such that it does not produce significant errors in the final
density.
Using the multigrid feature in ORCA is essentially very easy:
%method
FinalGrid 4 # grid to use for the final energy
# evaluation
end
The grid parameters of the final grid can also be individually specified:
%method
RadialGrid_fin GaussChebyshev
AngularGrid_fin Lebedev434
IntAcc_fin 5.0
NThetaMax_fin 0.7
GridPruning_fin 0
WeightScheme_fin Weight_Becke
HGridReduced_fin true
end
By default the multigrid feature is used and the final grid defaults to grid 4. If you do not want to use the
multigrid feature you have to specify:
%method
UseFinalGrid false
End
Or
! NoFinalGrid
CAUTION:
if you use a large grid (i.e. Grid=6) and forget to turn off the multigrid feature the energies you get
are only as accurate as Grid 4.
Post-SCF-GGA Option. For many molecules the cost of evaluating the XC energy and potential on a
three dimensional grid will dominate the overall computational cost. At the GGA level (functionals other than
the local density approximation, LSD and HFS, that incorporate the density gradient) it is often found that
the self-consistent electron distribution is rather similar to that obtained from a LSD calculation. However,
LSD calculations need a little bit less time because the derivatives of the basis functions at the grid points
are not needed. It may therefore be a sensible approach to calculate the self-consistent electron and spin
densities at the LSD level and then obtain the final total energy at the GGA level. This was advocated by the
ADF developers.
This feature is invoked in ORCA by using:
%method
PostSCFGGA true
end
What this does is to use the PWLDA local approximation in the SCF iterations and obtain the final energy at
whatever functional was chosen in the input file.
More recent test calculations reveal that the performance gain from using this feature is truly
minimal while the errors introduced in the final energies are large. Further use of this feature
is therefore strongly discouraged.
SpecialGrid Option. Sometimes you will like to increase the integration accuracy for some atoms that
need special care while it is not necessary to enlarge the grid generally. This situation for example arises
when the basis set contains very steep functions on a few atoms that are required in order to calculate core
properties such as isotropic hyperfine couplings. ORCA provides you with a basic mechanism to increase the
radial integration accuracy for a few atoms while maintaining the chosen grid for all others.
%method
# a maximum of 64 assignments can be made
# in = 0 : no changes are made
# in > 0 : the grid will be changed for all atoms with
# atomic number=in to IntAcc=an
# in < 0 : only the specific nth atom will have its
# IntAcc value changed to an
SpecialGridAtoms i1, i2, i3,...,in;
SpecialGridIntAcc a1,a2,a3,...,an;
end
9.2.2.3 Using the RI-J Approximation to the Coulomb Part
A very useful approximation that greatly speeds up DFT calculations unless the molecule gets very large
is the so called RI-approximation [223229]. RI stands for Resolution of the identity. In short, charge
distributions arising from products of basis functions are approximated by a linear combination of auxiliary
basis functions.
X
i (~r) j (~r) cij
k k (r) (9.8)
k
There are a variety of different possibilities to determine the expansion coefficients cij
k . A while ago Alml
of
and coworkers [247] have shown that for the approximation of electron repulsion integrals the best choice is
to minimize the residual repulsion 5 .
Define:
X
Rij i (~r) j (~r) cij
k k (~
r) (9.9)
k
and
Z Z
1
Tij = Rij (~r) Rij (~r) d3 rd3 r0 (9.10)
|~r ~r0 |
5
But note that the basic theory behind the method is known for a long time, at least since the late sixties have
methods similar to the RI approximation been used, mainly in the context of approximate ab initio methods
such as LEDO, PDDO and MADO but also in density functional theory in the mid and late seventies by Baerends,
Dunlap and others [223225, 229]
1
r) d3 rd3 r0 , leading to
RR
and determine Tij = Rij (~r) |~r~
r 0 | Rij (~
cij = V1 tij (9.11)
where:
tij

1
k = i j r12 k (9.12)

1
Vij = i r12 j (9.13)
Thus an ordinary two electron integral becomes:

1 X ij kl
i j r12 k l cp cq Vpq (9.14)
p,q
X X X
V1 tij V1 tkl

= Vpq pr r qs s
p,q r s
X
V1 tij tkl

= rs r s
(9.15)
r,s
and the total Coulomb energy becomes (P is the total density matrix):
XX
1
EJ = Pij Pkl i j r12 k l (9.16)
i,j k,l
XX X
V1 tij tkl

Pij Pkl rs r s
i,j k,l r,s
X X X
= V1 rs
Pij tij
r Pkl tkl
s (9.17)
r,s i,j k,l
| {z }| {z }
Xr Xs
In a similar way the Coulomb contribution to the Kohn-Sham matrix is calculated. There are substantial
advantages from this approximation: the quantities to be stored are the matrix V1 which depends only on
two indices and the three index auxiliary integrals tij
r . This leads to a tremendous data reduction and storage
requirements relative to a four index list of repulsion integrals. Furthermore the Coulomb energy and the
Kohn-Sham matrix contributions can be very quickly assembled by simple vector/matrix operations leading
to large time savings. This arises because each auxiliary basis function k (~r) appears in the expansion of
many charge distributions i (~r) j (~r). Unfortunately a similar strategy is less easily (or with less benefit)
applied to the Hartree-Fock exchange term. In addition, the two and three index electron repulsion integrals
are easier to compute than the four index integrals leading to further reductions in processing time.
If the auxiliary basis set {} is large enough, the approximation is also highly accurate. Since any DFT
procedure already has a certain, sometimes sizable, error from the noise in the numerical integration of the
XC part it might be argued that a similarly large error in the Coulomb part is perfectly acceptable without
affecting the overall accuracy of the calculation too much. Furthermore the errors introduced by the RI
method are usually much smaller than the errors in the calculation due to basis set incompleteness in the
first place. I therefore recommend the use of the RI procedure for pure DFs. However, one should probably
not directly mix absolute total energies obtained from RI and non-RI calculations because the error in the
total energy accumulates and will rise with increasing molecular size while the errors in the relative energies
will tend to cancel.
There are several choices for AUX basis sets described in the next section which depend on the choice of the
primary GTO basis used to expand the molecular orbitals6 .
In ORCA all that is needed to invoke the RI approximation is to type:
%method
RI on # do use the RI-J approximation
off # do not use the RI-J approximation
end
Recall:
If you use RI you must specify an auxiliary basis set in the %basis section. Do not rely on the program
to make an automatic choice for you.
9.2.2.4 The Split-RI-J Coulomb Approximation
There is an improved version of the RI-algorithm that has been implemented in version 2.2.09. This algorithm
yields the same Coulomb energy as the standard RI-algorithm but is significantly faster if the basis set
contains many high angular momentum functions (d-, f-, g-functions). For small basis sets there is virtually no
difference between the two algorithms except that Split-RI-J uses more memory than standard RI. However,
calculations with ca. 2000 basis functions need about 13 MB extra for Split-RI-J which is a trivial requirement
on present day hardware.
The Split-RI-J algorithm is invoked with:
! Split-RI-J
6
It probably should be noted that a slightly awkward step in the procedure is the inversion of the auxiliary integral
matrix V which can easily become very large. Matrix inversion is an O(N3 ) process such that for large molecules
this step takes some real time. However, in ORCA this is only done once during the calculation whereas other
programs that constrain the fit to also exactly reproduce the number of electrons need to perform a similar process
each iteration. Starting from Version 2.2.09 ORCA implements Cholesky decomposition in favor of matrix inversion
which removes any bottleneck concerning the solution of the linear equation system.
Split-RI-J is presently only available for SCF calculations. The gradient, coupled-perturbed Kohn-Sham and
TD-DFT modules will use the standard RI approximation instead.
NOTE:
The Split-RI-J algorithm is the default if RI is turned on via ! RI. If you do not want to use Split-RI-J
please insert the keyword ! NoSplit-RI-J
9.2.2.5 Using the RI Approximation for Hartree-Fock and Hybrid DFT (RIJONX)
The RI approximation can be used, although with less benefit, for hybrid DFT and Hartree-Fock (RHF
and UHF) calculations. In this case a different algorithm7 is used that allows a fair approximation to the
Hartree-Fock exchange matrix. It can be difficult to make this approximation highly accurate. It is, however,
usefully fast compared to direct SCF if the molecule is dense enough. There are special auxiliary basis sets
for this purpose (see section 6.3).
%method
RI on # do use the RI approximation
end
%basis
Aux "def2/JK"
end
NOTE: There has been little experimentation with this feature. It is provided on an experimental basis
here.
Another feature that was implemented is to use the RI method for the Coulomb term and the standard
treatment for the exchange term. This method is called RIJONX because the exchange term should tend
towards linear scaling for large molecules. You can use this feature for Hartree-Fock and hybrid DFT
calculation by using:
%method
RI on # do use the RI approximation
RIFlags 1 # ...but treat exchange exactly
end
# equivalently use the following keyword AND DONT FORGET

# TO ASSIGN AN AUXILIARY BASIS SET!
! RIJONX
The requirements for the auxiliary basis are the same as for the normal RI-J method.
7
This algorithm was described by Kendall and Fr
uchtl [226].
9.2.2.6 Using the RI Approximation for Hartree-Fock and Hybrid DFT (RIJCOSX)
The aim of this approximation is to efficiently compute the elements of exchange-type matrices:8
X
K = P (| ) (9.18)

where P is some kind of density-type matrix (not necessarily symmetric) and the two-electron integrals are
defined over the basis set {} by:
Z
1
(| ) = (r1 )(r1 )(r2 ) (r2 )r12 dr1 dr2 (9.19)
The approximation pursued here can be written as follows:
X X X
K Xg A (rg ) P Xg (9.20)
g
Here the index g refers to grid points rg and:
Xg = wg1/2 (rg ) (9.21)
Z
(r) (r)
A (rg ) = dr (9.22)
|r rg |
where wg denotes the grid weights. Thus, the first integration is carried out numerically and the second one
analytically. Note that this destroys the Hermitian character of the two-electron integrals.
Equation 9.20 is perhaps best evaluated in three steps:
F g = (PX) g (9.23)
X
Gg = A (rg )F g (9.24)

K = (XG+ ) (9.25)
As such the equations are very similar to the pseudo-spectral method extensively developed and discussed by
Friesner and co-workers since the mid 1980s and commercially available in the Jaguar quantum chemistry
package. The main difference at this point is that instead of Xg there appears a least-square fitting operator
Qg in Friesners formulation. Note that an analogue of the fitting procedure has also been implemented in
8
The theory of this approach together with all evaluations and implementation details is described in [104]. References
to earlier work can also be found there
ORCA, which however does not need specially optimized grids as in Friesners pseudospectral method. The
basic idea is to remove the grid errors within the basis set by fitting the numerical overlap to the analytical
one. Due to its nature, overlap fitting is supposed to work better with larger basis sets.
Given the exchange matrix, the exchange energy is given by (a sum over spin cases is left out here for
simplicity):
1X
EX = P K (P) (9.26)
2
Assuming that EX refers to the nonrelativistic, variational SCF exchange energy, the derivative with respect
to parameter can be re-arranged to the following form:
EX X X Fg
2 Gg (9.27)
g

with:
Fg X Xg
= wg1/2 P (9.28)

In this formulation, the gradient arises as a minor modification of the exchange matrix formation code.
In particular, the derivatives of the analytic integrals are not needed, merely the derivatives of the basis
functions on the grid.
In our implementation, we have defined grids of increasing quality. Higher and more accurate (and more
expensive) grids for the COSX approximation are chosen by increasing n from 1 to 9 in
! GridXn
From version 2.9, overlap fitting is used by default. This means that there might be some differences with
respect to previous versions. Firstly, smaller grids are used by default. The old default grid can be retained
by typing
! GridXOLD
If consistency with previous versions of ORCA is desired, overlap fitting can be removed using the keyword
!NoSFitting. Note that overlap fitting yields better results as the basis and/or the grid increases. It was
adapted as the default procedure with the smallest reliable grid, but it is compatible with all possible grids.
Grids tested in the overlap fitting paper are available as !GridXS1 and !GridXS2. [248] The latter grid is the
default.
For expert users, the grid parameters for the exchange grids can be even more finely controlled:
%method
IntAccX Acc1, Acc2, Acc3
GridX Ang1, Ang2, Ang3
XCorrection Corr1, Corr2, Corr3
UseFinalGridX true
end
There are three grids involved: the smallest grid (Acc1, Ang1) that is used for the initial SCF iterations, the
medium grid (Acc2, Ang2) that is used until the end of the SCF and the largest grid (Acc3, Ang3) that is
used for the final energy and the gradient evaluations. UseFinalGridX turns this last grid on or off. Accn
refers to the radial integration accuracy, GridX refers to the angular grid resolution and XCorrection (=0 or
1) refers to the method used for correcting the grid error. For 0 no one-center correction is applied, for
1 a one-center correction is computed analytically for the two-electron integrals. Note that in cases where
convergence difficulties arise, it is advisable to increase the smallest grid to a larger value, or even equal to
the medium grid using the Acc1, Ang1 parameters.
To modify the overlap fitting parameters
%method
UseSFitting false # equals to NoSFitting in the simple
# input. Default is true.
UseQGradFit true # uses the SCF fitting matrix for
# gradient calculations. Default
# is false.
end
Note that overlap fitting works for HF and MP2 gradients as well without specifying any additional keyword.
The UseQGradFit parameter merely uses the same fitting matrix for the gradients as for the energy calculation.
However, this does not save significant time, neither is it more accurate, therefore it is turned off by default.
9.2.2.7 Choice of the COSX Grid and Convergence Issues
Other then the !Gridn (where n is an integer number) keywords used to specify the DFT integration grids,
the grid structures used for HF exchange evaluation, called by the keywords !GridXn have also been discussed
in previous chapters. The purpose of this chapter is to give a more specific introduction into the properties of
COSX grids, and discuss some more common problems while using these.
As a start, it will be useful to relate the simple keywords GridXn, to the method block keywords detailed
above. Sometimes it is necessary to increase the grids using the method block parameters directly, as they
provide a more flexible control over the grid size. For some of the more common grids the simple keywords
and the equivalent method block parameters are given below
!GRIDX4
%method
GridX 1,1,3
IntAccX 3.34,4.01,4.34
end
!GRIDX5
%method
GridX 1,1,3
IntAccX 3.67,4.01,4.67
end
!GRIDX6
%method
GridX 1,2,3
IntAccX 3.67,4.01,4.34
end
!GRIDX7
%method
GridX 1,2,4
IntAccX 3.67,4.01,4.34
end
and for the default grid
!GRIDXS2
%method
GridX 1,1,2
IntAccX 3.34,3.67,4.01
end
Symptoms of convergence issues: Erratic convergence behavior, often starting from the first SCF step,
or possibly setting in at a later stage, producing crazy energy values, with megahartree jumps. If overlap
fitting is on, the following error message may also be encountered: Error in Cholesky inversion of numerical
overlap.
Convergence issues may arise when the chosen grid has difficulties in representing the basis set. This is the
grid equivalent of a linear dependence problem, discussed in 9.3.4. It should also be mentioned that the
grid related problem discussed here often goes hand in hand with basis set linear dependence, although not
necessarily. The most straightforward way of dealing with these is to increase the size of the integration grid.
This however, is not always desirable or possible, and in the remainder of this chapter some other methods
will be discussed.
One way to avoid the Cholesky inversion issue is to turn overlap fitting off (!NoSFitting), however, this
means that the numerical problems are still present, only they are ignored. Due to the fact that overlap
fitting operates with the numerical overlap, and its inverse, it is more sensitive to linear dependence issues,
and turning off the fitting procedure may lead to convergence. This may be a pragmatic, but by no means
clean solution, since it relies on the assumption that the numerical errors are small.
On the other hand, overlap fitting also gives a similar tool to deal with linear dependence issues as the
one discussed in 9.3.4 for basis sets. The eigenvalues of the numerical overlap can be inspected similarly,
and small values can be screened out. There is unfortunately no universal way to determine this screening
parameter, but see 9.3.4 for typical values.
The parameters influencing the method used for inversion and obtaining the fitting matrix are
%method
SFitInvertType Cholesky # Use Cholesky inversion. Default
Cholesky_Q # Cholesky + full Q matrix
Diag # Inversion via diagonalization
Diag_Q # Diag + full Q matrix
SInvThresh 1e-8 # inversion threshold for Diag and Diag_Q, default 1 e-8
end
By default, the inversion procedure proceeds through Cholesky decomposition as the fastest alternative.
Ideally the overlap matrix is non-singular, as long as the basis set is not linear dependent. For singular
matrices the Cholesky procedure will fail. It should be noted at this point that the numerical overlap can
become linear dependent even if the overlap of basis functions is not, and so a separate parameter will be
needed to take care of grid related issues. To achieve this, a diagonalization procedure (Diag) can be used
instead of Cholesky with the corresponding parameter to screen out eigenvectors belonging to eigenvalues
below a threshold (SInvThresh). For both Cholesky and diagonalization procedures a full Q approach
is also available (Cholesky Q and Diag Q), which corresponds to the use of a more accurate (untruncated)
fitting matrix.
9.2.2.8 Improved Analytical Evaluation of the Coulomb Term: Split-J
ORCA features a method that gives the exact Coulomb term at significantly reduced computational cost.
It can most profitably be applied to the case where no HF exchange is present. Thus if you use LDA or
GGA functionals and you do not want to apply the RI approximation (perhaps because you use a special
basis set for which no fit-set is available), the Split-J is an attractive alternative to the traditional evaluation.
The advantages of Split-J increase with the quality of the basis set used, i.e. if you have basis sets with
high-angular momentum functions split-J can be more effective by a factor of 2-5 compared to the traditional
evaluation. For smaller basis sets (i.e. SV(P) and the like) the advantages are smaller but still significant.
However, Split-J is also significantly slower than RI-J (but recall that Split-J is exact while RI-J is an
approximation).
A small job that uses the Split-J feature is shown below:

! RKS LSD TZVPP TightSCF Direct

%scf jmatrix 1 # turns on the Split-J feature
end
*int 0 1
C 0 0 0 0.00 0.0 0.00
O 1 0 0 1.20 0.0 0.00
H 1 2 0 1.10 120.0 0.00
H 1 2 3 1.10 120.0 180.00
*
9.2.2.9 The Standard Computational Levels
WARNING: The keywords below no longer change the basis set for the calculation. The
default basis set is def2-SVP and any changes to it must be specified explicitly! This makes
these keywords somewhat less helpful and their use is discouraged.
Standard calculation levels are defined below
Calculation-Level
DFT
Quick Quick Medium Good DFT
Parameter Acc Opt En-
DFT Opt Opt Opt Energy
ergy+
RunTyp Energy Opt Opt Opt Opt Energy Energy
Functional BP BP BP BP BP B3LYP B3LYP
1
RI On On on on on off off1
SCFConv Loose Tight Tight Tight Tight Normal Normal
GeoConv - Normal Normal Normal Tight - -
2
Grid 2/3 2/4 2/4 3/5 4/5 4/5 4/5
1 The RI approximation can be used in these calculations by using ! RIJONX in the input
2 The first and second number refer to the dual-grid feature of ORCA
You can easily override these defaults by simply typing the appropriate keyword that changes one of the settings
given above (for example if you want MediumOpt but with the PBE functional simply give ! MediumOpt
PBE).
9.2.2.10 Treatment of Dispersion Interactions with DFT-D3
Introduction
DFT-D3 is an atom-pairwise (atom-triplewise) dispersion correction which can be added to the KS-DFT
energies and gradient [2]:
EDFT-D3 = EKS-DFT + Edisp (9.29)

Edisp is then the sum of the two- and three-body contributions to the dispersion energy: Edisp = E (2) + E (3) .
The most important is the two-body term which is given at long range by:
X X CnAB
Edisp = sn n (9.30)
rAB
A<B n=6,8
CnAB denotes the averaged (isotropic) nth-order dispersion coefficient for atom pair AB, and rAB is their
internuclear distance. sn is a functional-dependent scaling factor (see below). In the general case, an adequate
damping function must be employed.
Damping Functions
In order to avoid near singularities for small rAB , the dispersion contribution needs to be damped at short
distances. One possible way is to use rational damping as proposed by Becke and Johnson [249251]:
X X CnAB
E (2) = sn n
(9.31)
A<B n=6,8
rAB + f (R0AB )n
with [251]
s
C8AB
R0AB = (9.32)
C6AB
and
f (R0AB ) = a1 R0AB + a2 . (9.33)
Damping the dispersion contribution to zero for short ranges (as in Ref. [2]) is also possible:
X X CnAB
E (2) = sn n fd,n (rAB ) (9.34)
rAB
A<B n=6,8
with
1
fd,n = rAB (9.35)
1 + 6( sr,n RAB
)n
0
Note that the R0AB used with this damping are from Ref. [2]. For more information on the supported damping
functions, see Ref [109]. In the ORCA program the dispersion correction with zero damping is invoked by
the keyword ! D3ZERO. The default is the recommended variant with Becke-Johnson damping and is invoked
by the keyword ! D3BJ or simply ! D3.
Three-body term
It is possible to calculate the three-body dispersion contributions with DFT-D3, according to
X C9ABC (3 cos a cos b cos c + 1)

E (3) = fd,(3) (rABC ), (9.36)
(rAB rBC rCA )3
A<B<C
where a , b and c are the internal angles of the triangle formed by rAB , rBC and rCA . The C9 coefficient is
approximated by
q
C9ABC C6AB C6AC C6BC . (9.37)
The three-body contribution has a small effect on medium-sized molecules and is damped according to
equation 9.36. The damping function fd,(3) (rABC ) is similar to the one shown in equation 9.35 with rABC
being the geometric mean of rAB , rBC and rCA . It can be used with both variants of the E (2) term. However,
the three-body term itself will always be calculated using the zero damping scheme. Adding the three-body
correction has proven to give quite accurate results for the thermochemistry of supramolecular systems [152].
In the ORCA program, the dispersion correction with zero damping and the three-body contribution is
invoked by the keyword ! D3ZERO ABC. Becke-Johnson damping for E (2) and zero damping for E (3) is invoked
by ! D3BJ ABC.
Options
Note that correcting Hartree-Fock (HF) is only parameterized with BJ-damping. For a constantly updated
list of supported functionals, check the website of the Grimme group [252]. If there is a functional on this
website that is parametrized, but the parameters are not implemented into the ORCA program yet, you can
specify the parameters manually as shown below (using the respective parameters from Ref. [252]). In the
same fashion, one could also use own parameters but this is not recommended.
Important: GGA and hybrid functionals should only be used with s6 = 1.0 to ensure asymptotically correct
behaviour. Double-hybrid functionals already account for parts of the dispersion interaction and should
therefore be used with s6 < 1.0. Within the %method block it is possible to change the parameters s6, a1,
s8 and a2 for the variant with Becke-Johnson damping.
! d3bj b2plyp
%method
D3S6 0.64
D3A1 0.3065
D3S8 0.9147
D3A2 5.0570
end
The variant with zero damping offers the parameters s6, rs6, s8 and 6 .
! d3zero blyp
%method
D3S6 1.0
D3RS6 1.094
D3S8 1.682
D3alpha6 14
end
If a geometry optimization is performed (! opt) then the program automatically calls the DFT-D3 gradient.
There are also special functional parameters, which were optimized for triple-zeta basis sets. This option is is
only available with zero damping and can be invoked by the keywords ! D3ZERO D3TZ. Preliminary results
in the SI of Ref. [2] indicate that results are slightly worse than with the default parameters and QZVP type
basis sets. This option should be carefully tested for future use in very large computations.
Example input files

In the following, we list some example input files. A computation using the DFT-D3 dispersion correction
with BJ-damping using the D3BJ keyword may look like this. As it is the default, the use of the keyword D3
is identical.
! pbe svp d3bj

* xyz 0 1
C 0.000000 0.000000 0.000000
O 0.000000 0.000000 1.400000
O 0.000000 0.000000 -1.400000
*
The output for the dispersion correction in the ORCA output will look like this:
-------------------------------------------------------------------------------
DFT DISPERSION CORRECTION
DFTD3 V2.1 Rev 6

USING Becke-Johnson damping
-------------------------------------------------------------------------------
The default PBE functional is recognized
Active option DFTDOPT ... 4
molecular C6(AA) [au] = 156.562480
DFT-D V3
parameters
s6 scaling factor : 1.0000
a1 scaling factor : 0.4289

ad hoc parameters k1-k3 : 16.0000 1.3333 -4.0000
Edisp/kcal,au: -0.563071585638 -0.000897311593

E6 /kcal : -0.390909076
E8 /kcal : -0.172162510
% E8 : 30.575598941
------------------------- ----------------
Dispersion correction -0.000897312
------------------------- ----------------
------------------------- --------------------
FINAL SINGLE POINT ENERGY -188.137098479095
------------------------- --------------------
Edisp is given as the Dispersion correction. It will be automatically included into the singlepoint energy.
As discussed above, the parameters s6, a1, s8 and a2 may be defined manually by:
! pbe svp d3bj

%method
D3S6 1.0
D3A1 0.4289
D3S8 0.7875
D3A2 4.4407
end
*xyz 0 1
C 0.000000 0.000000 0.000000
O 0.000000 0.000000 1.400000
O 0.000000 0.000000 -1.400000
*
The calculation of the same system with D3ZERO is invoked by:
! pbe svp d3zero

*xyz 0 1
C 0.000000 0.000000 0.000000
O 0.000000 0.000000 1.400000
O 0.000000 0.000000 -1.400000
*
In the same fashion as above, the s6, rs6, s8 and the exponent 6 can be defined by the user. The next
example shows this along with the call for the three-body correction using ABC:
! pbe svp d3zero abc

%method
D3S6 1.0000
D3RS6 1.2170
D3S8 0.7220
D3ALPHA6 14.0
end
*xyz 0 1
C 0.000000 0.000000 0.000000
O 0.000000 0.000000 1.400000
O 0.000000 0.000000 -1.400000
*
9.2.2.11 DFT Calculations with the Non-Local, Density Dependent Dispersion Correction:
DFT-NL
Accounting for the missing van der Waals (vdW, dispersion) forces in standard Kohn Sham Density
Functional Theory (DFT) has become essential in many studies of chemical and physical electronic structure
problems. Common approaches use atom pair-wise additive schemes such as the popular DFT-D3 [2, 109]
method, which is also available in ORCA by invoking the keyword ! D3 (for more information see section
9.2.2.10).
A different route is followed by the van der Waals Density Functional (vdW-DF) as pioneered by Langreth
and Lundquist [185]. These methods use as input to compute the Non-Local (NL) dispersion contribution
only the electron density. The recently developed vdW functional VV10 of Vydrov and Van Voorhis [253]
currently seems to be the most promising candidate for a general and accurate electronic structure method.
We use the term DFT-NL for any (hybrid)GGA density functional in combination with the non-local part of
the VV10 functional with an optimized parameter b, which will be defined below. The performance of these
methods has been evaluated in Ref. [254] using the GMTKN30 [197, 255, 256] database and the S66 set [257].
The performance of weak hydrogen bonds were evaluated in Ref. [258].
We recommend the DFT-D3 dispersion correction in general and the DFT-NL method for checking purposes
by single-point calculations. DFT-NL and DFT-D3 perform very similar but NL is to be preferred for metallic
systems or when the basic electronic structure changes significantly (e.g. oxidations or ionizations). Because
the NL correction increases the computational cost in particular of GGA treatments using RI significantly,
it is more useful in combination with hybrid functionals (and RI-JK or RIJCOSX techniques) where the
computational overhead is marginal if it is done non-self-consistently (see below).
The total exchange-correlation (XC) energy of VV10 type functionals is defined in eq. 9.38. It is composed
of standard exchange (X) and correlation (C) parts and the non-local (NL) term, which covers (mainly)
long-range dispersive energy:
DFT-NL (hybrid)GGA GGA VV10

EXC = EX + EC + EC-NL (9.38)
The NL term is given by the following double-integral:
Z Z
VV10 1 0 0 0
EC-NL = dr(r) + dr (r )(r, r ) (9.39)
2
where is the total electron density, and the definition of the kernel (r, r0 ) and is as follows (in a.u.):
(r, r0 ) = 2gg0 (g+g

3
0) g (r) = 0 (r) R2 + (r) R = |r r0 |
r 4 h i1/6 3 3/4
0 (r) = C (r)
(r) 1
(r) +
4
3 (r)
(r) = b 3 =

2 9 32 b2
In the original definition, the short-range attenuation parameter b appearing in and was fitted to the S22
set [259] of non-covalent interactions (b = 5.9 for the rPW86PBE GGA). The other parameter C = 0.0093
appearing in 0 , which determines the long-range behavior, was set to its original value. Other DFT-NL
functionals are constructed analogously. For a detailed discussion of the derivation of the formulas and their
physical meaning and basis see the references given and those given therein.
The defined energy of the non-local DFT-NL exchange-correlation functional can computed non-self-
consistently based on a converged SCF density but a self-consistent treatment is also possible. We take
B3LYP as an example.
In our implementation of the non-self-consistent B3LYP-NL functional, in the first step a self-consistent
B3LYP computation is performed. In the second step the optimized electron density from the B3LYP
computation is taken as input for the energy calculation of the non-local part. This procedure is invoked
by the combination of the keywords ! B3LYP NL. Use of the keywords ! B3LYP SCNL would request
a self-consistent treatment in which orbitals and density are optimized in the presence the full B3LYP-NL
exchange-correlation potential (see below). According to many test calculations, an SCNL treatment is rarely
necessary for normal energy evaluations.
The computation of the double-integral given in eq. 9.39 requires using an integration grid, just like for
normal exchange-correlation functionals. The grid size is used analogously to the regular grids available in
the ORCA program and can be specified by invoking the keyword ! vdwgridX, where X can be 1, 2, 3,
4, 5, 6, or 7. The default grid size is ! vdwgrid2. In test calculations we found, that going beyond grid
size of 3 is not necessary in normal applications.
In the following we compute the energy of an argon dimer at the distance of 3.76
A with the def2-TZVP basis
set using the B3LYP hybrid functional as an example with the non-self-consistent variant of the DFT-NL
dispersion correction. The original VV10 method is based on the rPW86PBE GGA.
! B3LYP NL
! def2-TZVP def2/JK RIJK grid4 vdwgrid2 nofinalgrid nososcf nopop
*xyz 0 1
Ar 0.0 0.0 0.0
Ar 0.0 0.0 3.76
*
The DFT-NL output for this example is shown below:
-------------------------------------------------------------------------------
DFT-NL dispersion correction
-------------------------------------------------------------------------------
SC Energy -1054.960659618
NL parameter b = 4.80
NL Energy 0.209416556
SC+NL Energy -1054.751243062
NL done in 0.3 sec
-------------------------------------------------------------------------------
----------------
TOTAL SCF ENERGY
----------------
Components:
NL Energy : 0.20941656 Eh 5.69851 eV
Here we find the B3LYP total energy (SC Energy) of -1054.9607 Eh, the parameter b = 4.8, the non-local
contribution (NL Energy) of 0.2094 Eh and the final total energy (SC+NL Energy) of -1054.7512 Eh,
which is the sum of the SC and NL energy. In the Components section the non-local-contribution is listed
separately (NL Energy) in order to be consistent with the ! SCNL option (see below).
In the current version of ORCA there are seven GGA and hybrid functionals available, which can be used
with the DFT-NL method. The GGA functionals are rPW86PBE (b = 5.9), BLYP (b = 4.0) and revPBE
(b = 3.7). The hybrid functionals are B3LYP (b = 4.8), B3PW91 (b = 4.5), revPBE0 (b = 4.3) and revPBE38
(b = 4.7). In addition, we also added the non-local term to Hartree-Fock (HF) with a parameter of b = 3.9.
The parameter C was not changed. All these functionals are ready to use by invoking the keyword ! DF
NL, where DF stands for the seven density functionals. Hartree-Fock is invoked with the keyword ! HF
NL.
All DFT-NL methods can be used for closed-shell and open-shell systems.
All density functionals that are available in ORCA (but for which no b parameter has been determined yet)
can also be used with the DFT-NL method by providing a value for the parameter b as shown here:
%method
NLb 5.0
End
For example, for the functional BP86 the parameter b has not been fitted yet and, therefore, using the
option ! BP86 NL will give an error, but by providing a parameter b (5.0 just as an example!) one can
use this functional in combination with the DFT-NL dispersion correction. Users who want to use such new
combinations might contact the Grimme group.
! BP86 NL
! def2-TZVP def2/J RI grid4 vdwgrid2 nofinalgrid nososcf nopop
%method
NLb 5.0
end
*xyz 0 1
Ar 0.0 0.0 0.0
Ar 0.0 0.0 3.76
*
In addition, for any of the already fitted functionals one can define a user specific value for the parameter b,
which might be different to the default value, therefore, overriding the default parameter b. For example, for
B3LYP the fitted parameter is b = 4.8 which is changed below to a more repulsive value of b = 5.2.
! B3LYP NL
%method
NLb 5.2
end
*xyz 0 1
Ar 0.0 0.0 0.0
Ar 0.0 0.0 3.76
*
The other parameter C = 0.0093 appearing in 0 may also be changed with the NLC keyword as shown in
the following example. Of course, both parameters b and C can be changed with the keywords NLb and
NLC at the same time.
! B3LYP NL
%method
NLC 0.0083
end
*xyz 0 1
Ar 0.0 0.0 0.0
Ar 0.0 0.0 3.76
*
Self-consistent computations with the DFT-NL dispersion correction
Self-consistent calculations with the DFT-NL dispersion correction are possible by invoking the keyword !
SCNL in combination with one of the available density functionals (rPW86PBE, revPBE, BLYP, B3LYP,
B3PW91, revPBE0 and revPBE38). However, as explained above any density functional can be used also in
combination with the SCNL keyword by providing a parameter b with the NLb keyword. Note, that due
to technical reasons self-consistent calculations are not possible with the Hartree-Fock method.
In the following example we use the B3LYP hybrid functional with the self-consistent DFT-NL variant:
! B3LYP SCNL
*xyz 0 1
Ar 0.0 0.0 0.0
Ar 0.0 0.0 3.76
*
The DFT-NL output of this example is shown below:
-------------------------------------------------------------------------------
Self-consistent DFT-NL dispersion correction
-------------------------------------------------------------------------------
NL parameter b = 4.80
--------------
SCF ITERATIONS
--------------
0 -1054.7512430617 0.000000000000 0.00166322 0.00007167 0.0007919 0.0000
1 -1054.7512472736 -0.000004211912 0.00077400 0.00002662 0.0005919 0.0000
2 -1054.7512477044 -0.000000430774 0.00004435 0.00000152 0.0000304 0.0000
**** Energy Check signals convergence ****
*****************************************************
* SUCCESS *
* SCF CONVERGED AFTER 3 CYCLES *
*****************************************************
----------------
TOTAL SCF ENERGY
----------------

Components:
NL Energy : 0.20940947 Eh 5.69832 eV
The procedure for the self-consistent computation is similar to the non-self-consistent one: In the first step
a self-consistent B3LYP run is performed and in the second step the converged electron density is used as
input for the self-consistent SCNL run, as shown in the output example above.
In the output the Total Energy is the energy of an exchange-correlation functional (B3LYP) plus the
non-local (NL) contribution. In the Components section the non-local contribution (NL Energy) is listed
separately in order to be consistent with the ! NL option (see above).
As can be seen from this example only two cycles of the self-consistent NL run were performed. The difference
in energy (5.0e-06) by comparing the self-consistent (-1054.751248) and non-self-consistent (-1054.751243)
computations is very small which seems to be typical also even for larger systems.
In the current version of ORCA analytical gradients for any DFT-NL dispersion correction are not available
and, therefore, geometry optimizations are not possible using analytical gradients. As mentioned above we
recommend to use DFT-D3 optimized geometries for single point DFT-NL calculations.
9.2.2.12 DFT and HF Calculations with the geometrical Counterpoise Correction: gCP
The central idea of the gCP correction [260] is to add in a semi-empirical fashion a correction EgCP to
the energies of molecular systems, in order to remove artificial overbinding effects from BSSE (see section
8.1.6). The correction uses atomic corrections and thus also yields the ability to correct for intramolecular
BSSE. The parametrization is constructed such that it approximates the Boys and Bernadi counterpoise
(CP) correction ECP in the intermolecular case
ECP EgCP , (9.40)
where e.g. for a complexation reaction A + B C our correction is given by
EgCP = EgCP (C) EgCP (A) EgCP (B). (9.41)
In practice, EgCP can simply be added to the HF/DFT energy
Etotal = EHF/DF T + EgCP , (9.42)
which is also done in ORCA. The FINAL SINGLE POINT ENERGY is the sum of the HF/DFT energy and the
gCP correction.
The central equation over all atoms N reads:
N X
X N
EgCP = emiss
a fdec (Rab ) , (9.43)
a b6=a
Table 9.4: Overview of parametrized basis sets. The level keyword in !GCP(level) is a com-
pound of HF or DFT and the basis set keyword. Valid examples are: !GCP(HF/MINIS),
!GCP(DFT/LANL), !GCP(HF/TZ), !GCP(DFT/631GD)
parametrized basis set HF DFT basis set keyword
MINIS yes yes MINIS
SV yes yes SV
6-31G(d) yes yes 631GD
6-31G(d) + LANL2DZ (Sc-Zn) no yes LANL
def2-SV(P) no yes SV(P)
def2-SV(P/h,c) no yes SVX or SV(P/H,C)
def2-SVP yes yes SVP
def2-TZVP yes yes TZ
where the energy emiss

a is a measure for the incompleteness of the chosen target basis set (that is typically
small), and fdec (Rab ) is a decay function that depends on the inter-atomic distance Rab .
The scaling factor is one out of 4 parameters needed for every method/basis set combination. More
details on this can be found in the original gCP paper [260].
Analytical gradients are available for geometry optimization. Due to its semi-empirical nature the correction
itself is calculated within seconds even for very large systems.
The correction can be invoked by using !GCP(level) keyword, where level is a compound of the method
(=HF or DFT) and the basis set. See table 9.4 for the available basis sets and the corresponding keyword.
For a B3LYP calculation using the def2-SV(P) basis set a typical input would be:
! B3LYP def2-SV(P) GCP(DFT/SV(P))

*xyzfile 0 1 example.xyz
The output states the level, the 4 parameters mentioned above and the correction itself:
------------------------------------------------------------------------------
g C P - geometrical counterpoise correction
------------------------------------------------------------------------------
Method: dft/sv(p)
Parameters: sigma eta alpha beta

0.2424 1.2371 0.6076 1.4078
Egcp: 0.0381115568 a.u.
------------------ -----------------
gCP correction 0.038111557
------------------ -----------------
------------------------- --------------------
------------------------- --------------------
There may be issued several warnings and notices: Elements of the 4th and higher period are treated as their
3rd period analog, i.e. Sn is treated as if it were Ge in terms of parameters. If this is the case a notice is
printed. If an element has no parameters it will give no contributions to the correction. Both cases can occur
at once:
! HF SV GCP(hf/minis)
*xyzfile 0 1 example.xyz
The corresponding output:
Method: hf/minis
** NOTE ** -> element sn will be treated as ge

** WARNING ** -> element ge has no parameters (no contribution)!
0.1290 1.1526 1.1549 1.1763
Egcp: 0.1593457998 a.u.
WARNING: Basis set mismatch between ORCA and otool_gcp:
ORCA: 280 gCP: 88
------------------ -----------------
gCP correction 0.159345800
------------------ -----------------
This is done for the user convenience. It enables the computation of molecules that has un-parametrized
elements. If only one or a few atoms in a larger molecule is treated inaccurate or not at all the error can
be expected to be small. In case of a mismatch between the ORCA basis set and the selected basis set for
gCP another warning is printed. This was done on purpose in the example above (GCP(hf/minis) instead of
GCP(hf/sv)). This is meant to be a safety check. However, a small deviation may cause only a minor error
(be careful nonetheless!). Use your own judgment if you want to use an unparametrized basis sets: Number
of basis functions and exponents should be very similar!
For debug issues, !LARGEPRINT will print additional information about parameters and individual atomic
contributions.
The gCP input can also be defined through the method section
%method
DoGCP true/false # turn gCP on/off
GCPMETHOD "method" # define method string for otool gcp, eg. dft/svp
GCP.D3MINIS true/false # use special DFT-D3 refit for HF/MINIS (default=true)
end
Further comments:
CP(HF/MINIS) sets automatically the refitted D3 parameter as proposed in the original gCP paper.
The gCP method is implemented via an external tool called otool gcp, which is based on the original Fortran
program used in the publication. Thus, the otool gcp binary can also be called directly via the command
line (otool gcp -h gives an overview of the options). It is also possible to read an external parameter file
($HOME/.gcppar) using !GCP(FILE). For further information, please look at the manual for the gcp program
as provided by Prof. S. Grimme9 . During the calculation some temporary output files are written by ORCA:
basename.gcp.in.tmp and basename.gcp.out.tmp will contain the input for otool gcp and its output.
It has been demonstrated that the popular combination of B3LYP with 6-31G(d) can be strongly im-
proved using DFT-D3 and gCP [261]. For convenience, the following compound keyword is defined
! B3LYP-gCP-D3/6-31G*. This equals the keywords: ! B3LYP 6-31G* D3BJ GCP(DF/631GD).
General Advices:
Small basis sets show not only a large BSSE, but general shortcomings. These effects are not always
clearly distinguishable.
If computationally affordable, large basis sets (triple- and higher) are always preferable for a given
system.
gCP makes only sense for somewhat large molecules
gCP should always be applied together with a dispersion correction for DFT: gCP-D3 is well tested,
but gCP-NL does also work well. (see sections 9.2.2.10 for DFT-D3 and 9.2.2.11 for DFT-NL)
9.2.2.13 HF-3c: Hartree-Fock with three corrections
HF-3c is a fast Hartree-Fock based method developed for computation of structures, vibrational frequencies
and non-covalent interaction energies in huge molecular systems [262]. The starting point for evaluating the
electronic energy is a standard HF calculation with a small Gaussian AO basis set. The used so-called MINIX
basis set consists of different sets of basis functions for different groups of atoms as shown in table 9.5.
Table 9.5: Composition of the MINIX basis set.

element basis
H-He, B-Ne MINIS
Li-Be MINIS+1(p)
Na-Mg MINIS+1(p)
Al-Ar MINIS+1(d)
K-Zn SV
Ga-Kr SVP
Rb-Rn def2-SVP with Stuttgart-Dresden ECPs
HF/MINIX
Three terms are added to correct the HF energy Etot in order to include London dispersion interactions,
to account for the BSSE and to correct for basis set deficiencies, i.e. overestimated bond lengths. The
corrected total energy is therefore calculated as
HF-3c HF/MINIX D3(BJ) gCP

Etot = Etot + Edisp + EBSSE + ESRB . (9.44)
D3(BJ)
The first correction term Edisp is the atom-pair wise London dispersion energy from the D3 dispersion
correction scheme [2] applying Becke-Johnson (BJ) damping [249251] (see section 9.2.2.10). The second term
gCP
EBSSE denotes the geometrical counterpoise (gCP) correction [260] to treat the BSSE (see section 9.2.2.12).
9
http://www.thch.uni-bonn.de/tc/
For the HF-3c method, the three usual D3 parameters s8 , a1 and a2 were re-fitted using reference interaction
energies of the complexes of the S66 test set [257]. This results in s8 = 0.8777, a1 = 0.4171 and a2 = 2.9149.
The parameter s6 was set to unity as usual to enforce the correct asymptotic limit and the gCP correction
was already applied in this fitting step.
The last term ESRB is a short-ranged correction to deal with basis set deficiencies which occur when using small
or minimal basis sets. It corrects for systematically overestimated covalent bond lengths for electronegative
elements and is calculated as a sum over all atom pairs:
Atoms
X Atoms
X 0,D3 3/4
ESRB = s (ZA ZB )3/2 exp((RAB ) RAB ) (9.45)
A B6=A
0,D3
Here, RAB are the default cut-off radii as determined ab initio for the D3 scheme [2] and ZA , ZB are the
nuclear charges. This correction is applied for all elements up to argon. The empirical fitting parameters
s = 0.03 and = 0.7 were determined to produce vanishing HF-3c total atomic forces for B3LYP-D3(BJ)/def2-
TZVPP equilibrium structures of 107 small organic molecules. More details can be found in the original
publication [262].
The HF-3c method can only be invoked with a simple keyword:
! HF-3c
! HF-3c is a compound keyword and equals ! HF MINIX D3BJ GCP(HF/MINIX) PATOM, hence no basis set
etc. has to be specified. The PATOM guess is selected since the grid construction for the default guess can take
more time than an actual SCF step. The guess can only be overwritten manually in the %method section.
The default mode for the integral handling is set to Conventional. The storing of the two-electron integrals
on disk or in memory if possible leads to large computational savings. In case one want to use the Direct
mode, this has to be specified in the %scf input section:
%scf
SCFmode Direct
end
The output gives the used parameters and the correction itself for D3 and gCP separately. As the SRB
correction is also calculated with the otool gcp, the results are given in the gCP output section. The Total
correction to HF/MINIX is the sum of all three corrections (D3, gCP and SRB) and FINAL SINGLE POINT
ENERGY is the total HF-3c energy as given in equation 9.44.
-------------------------------------------------------------------------------
DFT DISPERSION CORRECTION
DFTD3 V2.1 Rev 6

USING Becke-Johnson damping
-------------------------------------------------------------------------------
The default Hartree-Fock is recognized
Active option DFTDOPT ... 4
molecular C6(AA) [au] = 1689.256597
DFT-D V3
parameters
using HF/MINIX parameters
ad hoc parameters k1-k3 : 16.0000 1.3333 -4.0000
Edisp/kcal,au: -32.163184627631 -0.051255291794

E6 /kcal : -18.007221978
E8 /kcal : -14.155962649
% E8 : 44.012938437
------------------------- ----------------
Dispersion correction -0.051255292
------------------------- ----------------
------------------------------------------------------------------------------
g C P - geometrical counterpoise correction
------------------------------------------------------------------------------
Method: hf/minix

0.1290 1.1526 1.1549 1.1763
Egcp: 0.0723150636 a.u.
Ebas: -0.0636976872 a.u.
------------------ -----------------
gCP+bas correction 0.008617376
------------------ -----------------
---------------------------- ----------------
Total correction to HF/MINIX -0.042637915
---------------------------- ----------------
------------------------- --------------------
------------------------- --------------------
For the elements up to Xe, the default initial guess is a H

uckel guess. Beyond Xe, the guess mode is changed
to HCORE since no Huckel parameters for the respective ECP bases are available and other models are not
implemented at the moment. For calculations with only lighter elements and therefore no ECPs, the ECP
printouts in the output file can be ignored.
9.2.2.14 PBEh-3c: A PBE hybrid density functional with small AO basis set and two
corrections
PBEh-3c is a highly efficient electronic structure approach performing particularly well in the optimization
of geometries and for interaction energies of non-covalent complexes. [263] Here, a global hybrid variant of
the Perdew-Burke-Ernzerhof (PBE) functional with a relatively large amount of non-local Fock-exchange
(42%) is employed with a valence-double-zeta Gaussian AO basis set (def2-mSVP). Basis set superposition
errors (BSSE) and London dispersion effects are accounted for by the gCP and D3 schemes, respectively (see
above). The basis set is constructed such that:
Table 9.6: Composition of the def2-mSVP basis set.

element basis
H def2-SV(P) ( scaled by 1.2)
He def2-SVP(-p)
Li-Be,Na-Kr def2-SV(P)
B,Ne Ahlrichs DZ + Polarization from def2-SVP
C-F Ahlrichs DZ + Polarization from 6-31G*
Rb-Rn def2-SVP with Stuttgart-Dresden ECPs
For inter- and intramolecular BSSE the gCP expression from Eq. 9.43 is used but with a damping function
(similar to the zero-damping in Eq. 9.35). This damping improves the thermochemistry of the method
significantly compared with the non-damped version. London dispersion effects are accounted for by the
DFT-D3 (BJ-damping) scheme including the three-body term. Compared to the related HF-3c approach,
the PBEh-3c is somewhat more costly, however, it yields much better geometries. These are roughly of
MP2-quality (or even better for non-covalent structures) but may be computed at much lower cost. Due to
the moderate amount of non-local Fock exchange, the method is less prone to self-interaction errors (as in
GGAs) but still applicable in cases when Hartree-Fock fails (strongly correlated systems).
The PBEh-3c method may be invoked with the simple keyword:
! PBEh-3c
Identical to HF-3c, the default initial guess for all elements up to Xe is a Huckel guess. Beyond Xe, the
guess mode is changed to HCORE. For calculations with only lighter elements and therefore no ECPs, the ECP
printouts in the output file can be ignored.
Very recently, a new composite low-cost method for accurate thermochemistry, structures, and noncovalent
interactions specifically also for transition metal chemistry and other stronger correlated systems was
implemented. As it is based on the B97 GGA including D3 with thee-body contribution, a short range bond
length correction, and a modified, stripped-down triple- basis, def2-mTZVP, the computational cost of this
method termed B97-3c are between that of HF-3c and PBEh-3c (for large systems roughly two times more
expensive than HF-3c). It is invoked with a simple keyword analogously to the latter methods. A more
detailed description as well as the proper reference to cite will be added after the corresponding paper has
been published (approximately March 2017, S. Grimme et al.).
9.2.3 Semiempirical Methods
The present version of ORCA has inherited the capability of doing semiempirical calculations from the
earlier versions. A number of methods based on the neglect of differential overlap [264, 265] are currently
implemented for energies and analytic gradients (for geometry optimization).
%method
Method CNDO
INDO
NDDO
# for Method=CNDO
Version CNDO_1
CNDO_2
CNDO_S
# for Method=INDO
Version INDO_1
INDO_2
ZINDO_1
ZINDO_2
ZINDO_S
# for Method=NDDO
Version ZNDDO_1
ZNDDO_2
MNDO
AM1
PM3
end
The methods MNDO [266268], AM1 [269] and PM3 [270] are available for main group elements only and
arise from the work of the Dewar group. They have been optimized to reproduce molecular structure and
energetics. The older CNDO/1,2 and INDO/1,2 were developed by the Pople group [271279] and were
designed to roughly mimic minimal basis ab initio calculations. The methods of the Zerner group (ZINDO/1,2
and ZINDO/S) are closely related to the older methods but have been well parameterized for transition metals
too [280290]. ZINDO/1 (and less so ZINDO/2) are suitable for geometry optimization. ZINDO/S gives good
results for electronically excited states at moderate configuration interaction levels and is also successful for the
calculation of electron and spin distributions in large transition metal complexes [285290]. The ZNDDO/1,2
methods have been implemented into ORCA as straightforward extensions of the corresponding INDO
methods without changing any parameter. However, the methods benefit from the somewhat more accurate
representation of the Coulomb interaction within the NDDO approximation [291, 292]. The preliminary
experience with these methods is that they are better than the corresponding INDO methods for calculation
of transition metal complex structures but on the whole have also similar deficiencies.
The analytic gradients are available for all of these methods and can be used to produce reasonable molecular
structures at low computational cost or to get preliminary insight in the behavior of the system under
investigation10 .
There is also a mechanism for simplified input. Instead of giving values for Method and Version separately
you can also assign the value that would normally belong to Method to Version. The program will recognize
that and assign the correct values to both Method and Version.
%method
# shortcut to Method=NDDO; and Version=AM1;
Method AM1
end
If you want you can also combine semiempirical methods with MP2 (energies only). For example
use Method=AM1; and DoMP2=true; It is questionable if this makes the results of semiempirical
calculations any better but at least it is possible in ORCA.
You can change the built-in semiempirical parameters in a straightforward fashion. For example:
! RHF ZINDO/S TightSCF DIIS NoRICO NoMOPrint
%cis NRoots 20
MaxDim 3 # Davidson expansion space = MaxDim * NRoots
end
%ndoparas P[6,25] 20
P[6,26] 20
end
The %ndoparas block is there in order to let you input your favorite personal parameters. The molecular
parameters are set using INTFA (interaction factors);
%ndoparas INTFA[PP_PI] 0.585

# The interaction factors exist for
# ss_sigma
# sp_sigma
# sd_sigma
# pp_sigma
# pd_sigma
# dd_sigma
# pp_pi
# pd_pi
# dd_pi
10
However, do not try to use ZINDO/S (or CNDO/S) for structure optimizations - it does not make sense and will
lead to disastrous results because there is no accurate representation of nuclear repulsion in these methods.
# dd_delta
# the parameter entering the Coulomb integrals
# in INDO/S
FGAMMA 1.2
end
end
All atomic parameters are collected in an array P. The first index is the atomic number of the element whose
parameters you want to change. The second index identifies which parameter. The list of parameters follows
below. Most of them will only be interesting for expert users. The most commonly modified parameters are
the Betas (number 25 through 28). Note that most programs require a negative number here. In ORCA the
resonance integrals are defined in a way that makes the Betas positive.
# core integrals (in eV)

US 0
UP 1
UD 2
UF 3
# Basis set parameters (double-zeta for generality)
NSH 4 # number of shells for the element
NZS 5 # number of Slater type orbitals for the s shell
ZS1 6 # first exponent
ZS2 7 # second exponent
CS1 8 # first contraction coefficient
CS2 9 # second contraction coefficient
NZP 10 # number of Slater type orbitals for the p shell
ZP1 11 # ...
ZP2 12
CP1 13
CP2 14
NZD 15 # number of Slater type orbitals for the d shell
ZD1 16 # ...
ZD2 17
CD1 18
CD2 19
NZF 20 # number of Slater type orbitals for the f shell
ZF1 21 # ...
ZF2 22
CF1 23
CF2 24
# Resonance integral parameters (in eV)
BS 25 # s shell beta
BP 26 # p shell beta
BD 27 # d shell beta
BF 28 # f shell beta
# Number of electrons in the g.s.
NEL 29 # total number of electrons (integer)
NS 30 # fractional occupation number of the s shell
NP 31 # fractional occupation number of the p shell
ND 32 # fractional occupation number of the d shell
NF 33 # fractional occupation number of the f shell
# The one center repulsion (gamma) integrals (in eV)
GSS 34
GSP 35
GSD 36
GSF 37
GPP 38
GPD 39
GPF 40
GDD 41
GDF 42
GFF 43
# The Slater Condon parameters (in eV)
F2PP 44
F2PD 45
F2DD 46
F4DD 47
G1SP 48
G1PD 49
G2SD 50
G3PD 51
R1SPPD 52
R2SDPP 53
R2SDDD 54
# The nuclear repulsion parameters for Dewar type models
NR1 55
NR2 56
NR3 57
NR4 58
NR5 59
NR6 60
NR7 61
NR8 62
NR9 63
NR10 64
NR11 65
NR12 66
NR13 67
# The nuclear attraction/repulsion parameter for MNDO/d
RHO 68
# Spin orbit coupling parameters
SOCP 69 # SOC for the p shell
SOCD 70 # SOC for the d shell
SOCF 71 # SOC for the f shell
9.3 Choice of Basis Set
A fair number of reasonable basis sets is hardwired in the program as will be described in the next section. In
addition, whole basis sets can be read from a file, basis sets can be assigned for all atoms of a given type or,
at the highest resolution, basis sets can be assigned to individual atoms which is convenient if different parts
of the molecule are to be treated at different levels of accuracy. All hard wired basis sets were obtained from
the EMSL library [293] and the input format in ORCA is closely related to the GAMESS-US format.
As of ORCA version 4.0, the basis set handling has been

significantly modified!
Please check your basis sets very carefully!
9.3.1 Built-in Basis Sets
The basis set is specified in the block [BASIS]. The format is straightforward.
As of ORCA 4.0, the basis set name has to be put in quotation marks, as well as the basis set
name identifiers are the same as in the one-liner!
%basis
Basis "Def2-TZVP" # The orbital expansion basis set

Aux "Def2/J" # RI-J auxiliary basis set
AuxJK "Def2/JK" # RI-JK auxiliary basis set
AuxC "Def2-TZVP/C" # Auxiliary basis set for correlated
# calcualtions, e.g. RI-MP2
CABS "cc-pVDZ-F12-OptRI" # complementary auxiliary basis set
# for F12 calculations
Decontract false # if chosen "true" the program will

# decontract the basis set
end
WARNING:
9.3 Choice of Basis Set 327
ORCA uses pure d and f functions (5D and 7F instead of Cartesian 6D and 10F) for all basis sets.
This needs to be taken into account when results are compared with other programs, especially for
Pople-style basis sets that were optimized with Cartesian (6D) functions.
If you use Decontract: if your basis set arises from general contraction it will contain duplicate
primitives in several contractions and these will be removed such that only unique primitives remain
and there is no problem with redundancy.
A complete list of predefined basis sets and their availability is given in table 9.7.
Table 9.7: Basis sets availability

Keyword Availability Keyword Availability
Orbital basis sets (Basis) Orbital basis sets (Basis)
STO-3G HI cc-pVDZ HAr, CaKr
MINI HCa cc-pVTZ HAr, CaKr, Y, Ag, Au
MINIS HCa cc-pVQZ HAr, CaKr
MINIX HRn cc-pV5Z HAr, CaKr
MIDI HNa, AlK cc-pV6Z HHe, BeNe, AlAr
3-21G HCs aug-cc-pVDZ HAr, ScKr
3-21GSP HAr aug-cc-pVTZ HAr, ScKr, Ag, Au
4-22GSP HAr aug-cc-pVQZ HAr, ScKr
6-31G HZn aug-cc-pV5Z HAr, ScKr
6-31G* HZn aug-cc-pV6Z HHe, BNe, AlAr
m6-31G ScCu cc-pVD(+d)Z NaAr
m6-31G* ScCu cc-pVT(+d)Z NaAr
6-31G** HZn cc-pVQ(+d)Z NaAr
6-31G(d) HZn cc-pV5(+d)Z NaAr
6-31G(d,p) HZn aug-cc-pVTZ-J H, BF, AlCl, ScZn, Se
6-31G(2d) HZn cc-pCVDZ LiAr, Ca, GaKr
6-31G(2d,p) HZn cc-pCVTZ LiAr, Ca, GaKr
6-31G(2d,2p) HZn cc-pCVQZ LiAr, Ca, GaKr
6-31G(2df) HZn cc-pCV5Z LiAr, Ca, GaKr
6-31G(2df,2p) HZn cc-pCV6Z BNe, AlAr
6-31G(2df,2pd) HZn aug-cc-pCVDZ LiAr, GaKr
6-31+G* HZn aug-cc-pCVTZ LiAr, GaKr
6-31+G** HZn aug-cc-pCVQZ LiAr, GaKr
6-31+G(d) HZn aug-cc-pCV5Z LiAr, GaKr
6-31+G(d,p) HZn aug-cc-pCV6Z BNe, AlAr
6-31+G(2d) HZn cc-pwCVDZ LiAr, Ca, GaKr
6-31+G(2d,p) HZn cc-pwCVTZ LiAr, CaKr, Ag, Au
6-31+G(2d,2p) HZn cc-pwCVQZ LiAr, CaKr
6-31+G(2df) HZn cc-pwCV5Z LiAr, CaKr
6-31+G(2df,2p) HZn aug-cc-pwCVDZ LiAr, GaKr
6-31+G(2df,2pd) HZn aug-cc-pwCVTZ LiAr, ScKr, Ag, Au
6-31++G** HZn aug-cc-pwCVQZ LiAr, ScKr
6-31++G(d,p) HZn aug-cc-pwCV5Z LiAr, ScKr
6-31++G(2d,p) HZn cc-pVDZ-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-31++G(2d,2p) HZn cc-pVTZ-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-31++G(2df,2p) HZn cc-pVQZ-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-31++G(2df,2pd) HZn cc-pV5Z-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-311G HBr aug-cc-pVDZ-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-311G* HBr aug-cc-pVTZ-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-311G** HBr aug-cc-pVQZ-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-311G(d) HBr aug-cc-pV5Z-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-311G(d,p) HBr cc-pCVDZ-PP Ca, Sr, Ba, Ra
6-311G(2d) HBr cc-pCVTZ-PP Ca, Sr, Ba, Ra
6-311G(2d,p) HBr cc-pCVQZ-PP Ca, Sr, Ba, Ra

6-311G(2d,2p) HBr cc-pCV5Z-PP Ca, Sr, Ba, Ra
6-311G(2df) HBr aug-cc-pCVDZ-PP Ca, Sr, Ba, Ra
6-311G(2df,2p) HBr aug-cc-pCVTZ-PP Ca, Sr, Ba, Ra
6-311G(2df,2pd) HBr aug-cc-pCVQZ-PP Ca, Sr, Ba, Ra
6-311G(3df) HBr aug-cc-pCV5Z-PP Ca, Sr, Ba, Ra
6-311G(3df,3pd) HBr cc-pwCVDZ-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-311+G* HBr cc-pwCVTZ-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-311+G** HBr cc-pwCVQZ-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-311+G(d) HBr cc-pwCV5Z-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-311+G(d,p) HBr aug-cc-pwCVDZ-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-311+G(2d) HBr aug-cc-pwCVTZ-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-311+G(2d,p) HBr aug-cc-pwCVQZ-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-311+G(2d,2p) HBr aug-cc-pwCV5Z-PP Ca, CuKr, SrXe, Ba, HfRn, Ra
6-311+G(2df) HBr cc-pVDZ-DK HAr, ScKr
6-311+G(2df,2p) HBr cc-pVTZ-DK HAr, ScKr, YXe, HfRn
6-311+G(2df,2pd) HBr cc-pVQZ-DK HAr, ScKr, InXe, TlRn
6-311+G(3df) HBr cc-pV5Z-DK HAr, ScKr
6-311+G(3df,3pd) HBr aug-cc-pVDZ-DK HAr, ScKr
6-311++G** HBr aug-cc-pVTZ-DK HAr, ScKr, YXe, HfRn
6-311++G(d,p) HBr aug-cc-pVQZ-DK HAr, ScKr, InXe, TlRn
6-311++G(2d,p) HBr aug-cc-pV5Z-DK HAr, ScKr
6-311++G(2d,2p) HBr cc-pwCVDZ-DK LiBe, NaMg, CaZn
6-311++G(2df,2p) HBr cc-pwCVTZ-DK LiBe, NaMg, CaZn, YXe, HfRn
6-311++G(2df,2pd) HBr cc-pwCVQZ-DK LiBe, NaMg, CaZn, InXe, TlRn
6-311++G(3df,3pd) HBr cc-pwCV5Z-DK LiBe, NaMg, CaZn
SV HKr aug-cc-pwCVDZ-DK LiBe, NaMg, ScZn
SV(P) HKr aug-cc-pwCVTZ-DK LiBe, NaMg, ScZn, YXe, HfRn
SVP HKr aug-cc-pwCVQZ-DK LiBe, NaMg, ScZn, InXe, TlRn
TZV HKr aug-cc-pwCV5Z-DK LiBe, NaMg, ScZn
TZV(P) HKr Partridge-1 H, LiSr
TZVP HKr Partridge-2 H, LiKr
TZVPP HKr Partridge-3 H, LiZn
QZVP HKr Partridge-4 ScZn
QZVPP HKr ANO-SZ HAr, ScZn
DKH-SV(P) HKr LANL08 NaLa, HfBi
DKH-SVP HKr LANL08(f) ScCu, YAg, LaLa, HfAu
DKH-TZV(P) HKr LANL2DZ H, LiLa, HfBi, UPu
DKH-TZVP HKr LANL2TZ ScZn, YCd, LaLa, HfHg
DKH-TZVPP HKr LANL2TZ(f) ScCu, YAg, LaLa, HfAu
DKH-QZVP HKr Sapporo-DZP-2012 HXe
DKH-QZVPP HKr Sapporo-TZP-2012 HXe
ZORA-SV(P) HKr Sapporo-QZP-2012 HXe
ZORA-SVP HKr Sapporo-DKH3-DZP-2012 KRn
ZORA-TZV(P) HKr Sapporo-DKH3-TZP-2012 KRn
ZORA-TZVP HKr Sapporo-DKH3-QZP-2012 KRn
ZORA-TZVPP HKr SARC-DKH-SVP HfHg
ZORA-QZVP HKr SARC-DKH-TZVP XeRn, AcLr
ZORA-QZVPP HKr SARC-DKH-TZVPP XeRn, AcLr
def2-mSVP HRn SARC-ZORA-SVP HfHg
def2-mTZVP HRn SARC-ZORA-TZVP XeRn, AcLr
def2-SV(P) HRn SARC-ZORA-TZVPP XeRn, AcLr
def2-SVP HRn SARC2-DKH-QZV LaLu
def2-TZVP(-f) HRn SARC2-DKH-QZVP LaLu
def2-TZVP HRn SARC2-ZORA-QZV LaLu
def2-TZVPP HRn SARC2-ZORA-QZVP LaLu
def2-QZVP HRn D95 H, LiLi, BNe, AlCl
def2-QZVPP HRn D95p H, LiLi, BNe, AlCl
def2-SVPD HLa, HfRn EPR-II H, BF

def2-TZVPD HLa, HfRn EPR-III H, BF
def2-TZVPPD HLa, HfRn IGLO-II H, BF, AlCl
def2-QZVPD HLa, HfRn IGLO-III H, BF, AlCl
def2-QZVPPD HLa, HfRn UGBS HTh, PuAm, CfLr
DKH-def2-SV(P) HKr CP ScZn
DKH-def2-SVP HKr CP(PPP) ScZn
DKH-def2-TZVP(-f) HKr Wachters+f ScCu
DKH-def2-TZVP HKr cc-pVDZ-F12 HAr
DKH-def2-TZVPP HKr cc-pVTZ-F12 HAr
DKH-def2-QZVPP HKr cc-pVQZ-F12 HAr
ZORA-def2-SV(P) HKr cc-pVDZ-PP-F12 GaKr, InXe, TlRn
ZORA-def2-SVP HKr cc-pVTZ-PP-F12 GaKr, InXe, TlRn
ZORA-def2-TZVP(-f) HKr cc-pVQZ-PP-F12 GaKr, InXe, TlRn
ZORA-def2-TZVP HKr cc-pCVDZ-F12 LiAr
ZORA-def2-TZVPP HKr cc-pCVTZ-F12 LiAr
ZORA-def2-QZVPP HKr cc-pCVQZ-F12 LiAr
ma-def2-mSVP HRn Coulomb-fitting auxiliary basis sets (AuxJ)
ma-def2-SV(P) HRn def2/J HRn
ma-def2-SVP HRn def2-mTZVP/J HRn
ma-def2-TZVP(-f) HRn SARC/J HRn, AcNo
ma-def2-TZVP HRn Coulomb and exchange-fitting auxiliary basis sets (AuxJK)
ma-def2-TZVPP HRn def2/JK HBa, HfRn
ma-def2-QZVP HRn def2/JKsmall HRa, ThLr
ma-def2-QZVPP HRn cc-pVTZ/JK H, BF, AlCl, GaBr
ma-DKH-def2-SV(P) HKr cc-pVQZ/JK H, BF, AlCl, GaBr
ma-DKH-def2-SVP HKr cc-pV5Z/JK H, BF, AlCl, GaBr
ma-DKH-def2-TZVP(-f) HKr aug-cc-pVTZ/JK H, BF, AlCl, GaBr
ma-DKH-def2-TZVP HKr aug-cc-pVQZ/JK H, BF, AlCl, GaBr
ma-DKH-def2-TZVPP HKr aug-cc-pV5Z/JK H, BF, AlCl, GaBr
ma-DKH-def2-QZVPP HKr SARC2-DKH-QZV/JK LaLu
ma-ZORA-def2-SV(P) HKr SARC2-DKH-QZVP/JK LaLu
ma-ZORA-def2-SVP HKr SARC2-ZORA-QZV/JK LaLu
ma-ZORA-def2-TZVP(-f) HKr SARC2-ZORA-QZVP/JK LaLu
ma-ZORA-def2-TZVP HKr Auxiliary basis sets for correlated methods (AuxC)
ma-ZORA-def2-TZVPP HKr def2-SVP/C HLa, HfRn
ma-ZORA-def2-QZVPP HKr def2-TZVP/C HLa, HfRn
old-SV HI def2-TZVPD/C HLa, HfRn
old-SV(P) HI def2-TZVPP/C HLa, HfRn
old-SVP HI def2-TZVPPD/C HLa, HfRn
old-TZV HI def2-QZVPP/C HLa, HfRn
old-TZV(P) HI def2-QZVPPD/C HLa, HfRn
old-TZVP HI cc-pVDZ/C HAr, GaKr
old-TZVPP HI cc-pVTZ/C HAr, ScKr
old-DKH-SV(P) HI cc-pVQZ/C HAr, ScKr
old-DKH-SVP HI cc-pV5Z/C HAr, GaKr
old-DKH-TZV(P) HI cc-pV6Z/C HHe, BNe, AlAr
old-DKH-TZVP HI aug-cc-pVDZ/C HHe, BeNe, MgAr, GaKr
old-DKH-TZVPP HI aug-cc-pVTZ/C HHe, BeNe, MgAr, ScKr
old-ZORA-SV(P) HI aug-cc-pVQZ/C HHe, BeNe, MgAr, ScKr
old-ZORA-SVP HI aug-cc-pV5Z/C HNe, AlAr, GaKr
old-ZORA-TZV(P) HI aug-cc-pV6Z/C HHe, BNe, AlAr
old-ZORA-TZVP HI cc-pwCVDZ/C BNe, AlAr, GaKr
old-ZORA-TZVPP HI cc-pwCVTZ/C BNe, AlAr, ScKr
ANO-pVDZ HAr, ScZn cc-pwCVQZ/C BNe, AlAr, GaKr
ANO-pVTZ HAr, ScZn cc-pwCV5Z/C LiNe, AlAr
ANO-pVQZ HAr, ScZn aug-cc-pwCVDZ/C BNe, AlAr, GaKr
ANO-pV5Z HAr, ScZn aug-cc-pwCVTZ/C BNe, AlAr, ScKr
ANO-pV6Z HAr, ScZn aug-cc-pwCVQZ/C BNe, AlAr, GaKr

aug-ANO-pVDZ HAr, ScZn aug-cc-pwCV5Z/C LiNe, AlAr
aug-ANO-pVTZ HAr, ScZn cc-pVDZ-PP/C CuKr, YXe, HfRn
aug-ANO-pVQZ HAr, ScZn cc-pVTZ-PP/C CuKr, YXe, HfRn
aug-ANO-pV5Z HAr, ScZn cc-pVQZ-PP/C CuKr, YXe, HfRn
saug-ANO-pVDZ HAr, ScZn aug-cc-pVDZ-PP/C CuKr, YXe, HfRn
saug-ANO-pVTZ HAr, ScZn aug-cc-pVTZ-PP/C CuKr, YXe, HfRn
saug-ANO-pVQZ HAr, ScZn aug-cc-pVQZ-PP/C CuKr, YXe, HfRn
saug-ANO-pV5Z HAr, ScZn cc-pwCVDZ-PP/C CuKr, YXe, HfRn
ANO-RCC-DZP HCm cc-pwCVTZ-PP/C CuKr, YXe, HfRn
ANO-RCC-TZP HCm cc-pwCVQZ-PP/C CuKr, YXe, HfRn
ANO-RCC-QZP HCm aug-cc-pwCVDZ-PP/C CuKr, YXe, HfRn
ANO-RCC-Full HCm aug-cc-pwCVTZ-PP/C CuKr, YXe, HfRn
pc-0 HCa, GaKr aug-cc-pwCVQZ-PP/C CuKr, YXe, HfRn
pc-1 HKr cc-pVDZ-F11-MP2Fit HAr
pc-2 HKr cc-pVTZ-F11-MP2Fit HAr
pc-3 HKr cc-pVQZ-F11-MP2Fit HAr
pc-4 HKr cc-pVDZ-PP-F11-MP2Fit GaKr, InXe, TlRn
aug-pc-0 HCa, GaKr cc-pVTZ-PP-F11-MP2Fit GaKr, InXe, TlRn
aug-pc-1 HKr cc-pVQZ-PP-F11-MP2Fit GaKr, InXe, TlRn
aug-pc-2 HKr cc-pCVDZ-F11-MP2Fit LiAr
aug-pc-3 HKr cc-pCVTZ-F11-MP2Fit LiAr
aug-pc-4 HKr cc-pCVQZ-F11-MP2Fit LiAr
pcJ-0 HHe, BNe, AlAr Complementary auxiliary basis sets for F12 calculations (CABS)
pcJ-1 HHe, BNe, AlAr cc-pVDZ-F11-CABS H, BNe, AlAr
pcJ-2 HHe, BNe, AlAr cc-pVTZ-F11-CABS H, BNe, AlAr
pcJ-3 HHe, BNe, AlAr cc-pVQZ-F11-CABS H, BNe, AlAr
pcJ-4 HHe, BNe, AlAr cc-pVDZ-F11-OptRI HAr
aug-pcJ-0 HHe, BNe, AlAr cc-pVTZ-F11-OptRI HAr
aug-pcJ-1 HHe, BNe, AlAr cc-pVQZ-F11-OptRI HAr
aug-pcJ-2 HHe, BNe, AlAr cc-pVDZ-PP-F11-OptRI GaKr, InXe, TlRn
aug-pcJ-3 HHe, BNe, AlAr cc-pVTZ-PP-F11-OptRI GaKr, InXe, TlRn
aug-pcJ-4 HHe, BNe, AlAr cc-pVQZ-PP-F11-OptRI GaKr, InXe, TlRn
pcseg-0 HKr aug-cc-pVDZ-PP-OptRI CuZn, AgCd, AuHg
pcseg-1 HKr aug-cc-pVTZ-PP-OptRI CuZn, AgCd, AuHg
pcseg-2 HKr aug-cc-pVQZ-PP-OptRI CuZn, AgCd, AuHg
pcseg-3 HKr aug-cc-pV5Z-PP-OptRI CuZn, AgCd, AuHg
pcseg-4 HKr cc-pCVDZ-F11-OptRI LiAr
aug-pcseg-0 HKr cc-pCVTZ-F11-OptRI LiAr
aug-pcseg-1 HKr cc-pCVQZ-F11-OptRI LiAr
aug-pcseg-2 HKr aug-cc-pwCVDZ-PP-OptRI CuZn, AgCd, AuHg
aug-pcseg-3 HKr aug-cc-pwCVTZ-PP-OptRI CuZn, AgCd, AuHg
aug-pcseg-4 HKr aug-cc-pwCVQZ-PP-OptRI CuZn, AgCd, AuHg
pcSseg-0 HKr aug-cc-pwCV5Z-PP-OptRI CuZn, AgCd, AuHg
pcSseg-1 HKr
pcSseg-2 HKr
pcSseg-3 HKr
pcSseg-4 HKr
aug-pcSseg-0 HKr
aug-pcSseg-1 HKr
aug-pcSseg-2 HKr
aug-pcSseg-3 HKr
aug-pcSseg-4 HKr
A note on RI and auxiliary basis sets: one thing that is certainly feasible and reasonable if you do not want
to depend on the RI approximation is to converge a RI-J calculation and then take the resulting orbitals as
initial guess for a calculation with exact Coulomb term. This should converge within a few cycles and the
total execution time should still be lower than just converging the calculation directly with exact Coulomb
treatment.
AutoAux. If no auxiliary basis set is available for your chosen orbital basis set, one can be generated
automatically by ORCA using the keyword AutoAux. This is specified as any other fitting basis set: as a
separate keyword in the simple input line or as a value to the AUX variable in the %basis block (AutoAux can
also be assigned to individual elements or atoms - see sections 9.3.2 and 9.3.3). The generated basis sets can
be used for Coulomb, exchange and correlation fitting and are as accurate as the optimized auxiliary basis
sets at the cost of being up to twice as large. The exact generation procedure is described elsewhere [294] but
notably it has been significantly altered since ORCA 3.1 and will not produce the same results!
For compatibility, the old version is still accessible via the setting OldAutoAux true in the %basis block.
Some additional settings for AutoAux are given below with their default values.
%basis
Aux "AutoAux" # Use AutoAux to generate the fitting basis set
AutoAuxSize 1-3 # 1 (default) increases the maximal exponent
# for the shells with low angular momenta.
# 2 increases the maximal exponent for all shells
# 3 directly uses the primitives and produces
# the largest fitting basis
AutoAuxLmax false # If true increase the maximal angular momentum of
# the fitting basis set to the highest value
# permitted by ORCA and by the orbital basis set.
OldAutoAux false # If true selects the ORCA 3.1 generation procedure
# Only change the defaults below if you know what you are doing
AutoAuxF[0] 20.0 # The factor to increase the maximal s-exponent

AutoAuxF[1] 7.0 # Same for the p-shell
AutoAuxF[2] 4.0 # Same for the d-shell
AutoAuxF[3] 4.0 # Same for the f-shell
AutoAuxF[4] 3.5 # Same for the g-shell
AutoAuxF[5] 2.5 # Same for the h-shell
AutoAuxF[6] 2.0 # Same for the i-shell
AutoAuxF[7] 2.0 # Same for the j-shell
AutoAuxB[0] 1.8 # Even-tempered expansion factor for the s-shell
AutoAuxB[1] 2.0 # Same for the p-shell
AutoAuxB[2] 2.2 # Same for the d-shell
AutoAuxB[3] 2.2 # Same for the f-shell
AutoAuxB[4] 2.2 # Same for the g-shell
AutoAuxB[5] 2.3 # Same for the h-shell
AutoAuxB[6] 3.0 # Same for the i-shell
AutoAuxB[7] 3.0 # Same for the j-shell
AutoAuxTightB true # Only use B[l] for shells with high l and B[0] for the rest
end
9.3.2 Assigning or Adding Basis Functions to an Element
In order to assign a new basis set to a given element, use:
%basis
NewGTO 8 # new basis for oxygen.
# NewGTO O # works as well
S 3
1 910.10034975 0.03280967
2 137.19711335 0.23422391
3 30.85279077 0.81490980
S 2
1 1.72885887 0.27389659
2 0.39954770 0.79112437
P 1
1 8.35065975 1.00000000
end
end
Note that for simplicity and consistency the input format is the same as that used in the basis set files. In
this format the first line carries first the angular momentum of the shell to be added (s, p, d, f, g, h, i, j) and
the number of primitives. Then for each primitive one line follows which has (a) the index of the primitive
(1, 2, 3, . . . ) (b) the exponent of the primitive and (c) the contraction coefficient (unnormalized). Note
that ORCA always uses spherical harmonic gaussian functions. L-shells (not to be confused with angular
momentum equal to 9) can only be dealt with as separate s- and p-shells. There also is the possibility to
include a SCALE X statement after the number of primitives in the first line to indicate that the basis function
exponents should be scaled.
In order to add basis functions to the basis of a given element (for example because you do not like the
standard polarization functions) use AddGTOs instead or NewGTOs. In NewGTOs or AddGTOs you can also use
the nicknames of internally stored basis sets. An example is:
%basis
NewGTO 8 # new basis for oxygen
"6-31G"
D 1
1 0.4 1.0
end
end
In this example the 6-31G basis is assigned to oxygen and in addition a polarization function with exponent
0.4 is added to the oxygen basis.
A similar mechanism was established for the auxiliary basis sets in RI calculations:
%basis
NewAuxGTO 8 # new auxiliary basis for oxygen
s 1
1 350 1.0
... etc
end
AddAuxGTO 8 # add a shell to the auxiliary basis for
# oxygen
D 1
1 0.8 1.0
end
end
9.3.3 Assigning or Adding Basis Functions to Individual Atoms
Sometimes you may want to not treat all carbon atoms with the same basis set but to assign a specific basis
set to a specific atom in the molecules. This is also possible in ORCA and takes place in the section [COORDS].
The format is the same as described above. An example may help to make things clear:
%coords
CTyp = Internal; # choose internal coordinates
Units = Angs; # choose Angstr
om units
Charge = 0; # total charge
Mult = 1; # spin multiplicity
Coords # start coordinate assignments
C(1) 0 0 0 0.00 0.0 0.00
AddGTO
D 1
1 1.0 1.0
end;
O(2) 1 0 0 1.13 0.0 0.00
NewGTO
"6-311G"
D 1
1 1.2 1.0
end;
end;
end
In this example an extra d-shell with exponent 1.0 is added to the first carbon atom and the basis for the
oxygen atom is changed to 6-311G with an extra d-function of exponent 1.2 added.
Analogously, AUX basis functions can be assigned or added to individual atoms using the keywords NewAuxGTO
and AddAuxGTO, respectively.
A note on the use of AutoAux: if you change the basis set on a given atom and want to generate a fitting
basis, you have to specify it again in the COORDS section, even if AutoAux is already present in the simple
input line or in the %basis block. For example:
! Def2-SVP Def2/JK
%basis
NewAuxGTO H
"AutoAux"
end
end
%coords
Coords
O 0.00 0.00 0.00
H -0.25 0.93 0.00
H 0.96 0.00 0.00
AddGTO
P 1
1 1.6 1.0
D 1
1 1.0 1.0
end
NewAuxGTO
"AutoAux"
end
end
end
Here the oxygen atom is assigned the Def2-SVP basis and the Def2/JK fitting basis, the first hydrogen atom
is assigned the Def2-SVP basis and an automatically generated fitting basis and the second hydrogen atom is
assigned the Def2-SVP basis with two additional polarization functions and a larger automatically generated
fitting basis that accounts for these functions.
TIP
When assigning custom basis sets it is always a good idea to print the basis set information (%output
print[p basis] 2 end) and check that everything is correct.
9.3.4 Linear Dependence
The previous sections describe the assessment of a desired molecular basis set from appropriately parametrized
functions at various locations within the molecule (normally centered on atoms). The parametrization of
these functions is such that the chance for redundancy is minimal. Since however, one is limited to work
with finite numerical precision, and furthermore these parameters also depend on the molecular geometry,
redundancies cannot be completely eliminated in advance. Redundancy means that the subspace spanned by
the given basis functions at given values of parameters (including geometry), can be identically spanned by a
smaller number of linear independent basis functions. Linear dependent (redundant) function sets however
may cause numerical instabilities. Linear dependence is normally identified by searching for zero eigenvalues
of the overlap matrix. Note that the inverse of the overlap (or related matrices) are used for orthogonalization
purposes, and it follows that if near zero eigenvalues are not treated properly, the inverse becomes ill-defined,
and the SCF procedure numerically unstable.
From the previous discussion, it is evident that the crucial parameter for curing linear dependence is the
threshold below which an overlap eigenvalue is considered zero. This parameter may be changed using the
following keyword
%scf
sthresh 1e-7 # default 1e-8
end
Although there is no strict limit for the value of the above parameter, it should reasonably be somewhere
between 1e-5 and 1e-8 (the latter is the default). One may get away with 1e-9 or perhaps even lower without
convergence problem, but there is a risk that the result is contaminated with noise caused by the near zero
vectors. In difficult cases, an 1e-7 threshold was often found to work smoothly, and above that one risks
throwing away more and more functions, which also influence comparability of results with other calculations.
To monitor the behaviour of the small eigenvalues, one should look for the following block in the output
Diagonalization of the overlap matrix:

Smallest eigenvalue ... -1.340e-17
Time for diagonalization ... 0.313 sec
Threshold for overlap eigenvalues ... 1.000e-08
Number of eigenvalues below threshold ... 1
Time for construction of square roots ... 0.073 sec
Total time needed ... 0.387 sec
Here, the smallest eigenvalue is printed, along with the currently used overlap threshold, and the number of
functions below this (which will be dropped). It is a recommended consistency check to look for an equal
number of zero entries among orbital energies once the SCF procedure converged. Note that for functions
belonging to zero eigenvalues no level shifts are applied!
9.3.5 Reading Orbital and Auxiliary Basis Sets from a File
By using the variables GTOName and GTOAuxName a basis set can be read from an ASCII file11 . The format is
that used for GAMESS-US in the EMSL library [293].
%basis
# read an externally specified orbital basis
GTOName = "MyBasis.bas";
11
In order to not confuse the program it is recommended to assign some value to Basis and Aux and then overrule
this default with the external basis set. The externally specified basis set then takes priority.
# read an externally specified basis for RI

# calculations
GTOAuxName= "MyAuxBasis.bas";
# for STO basis
STOName= "MySTOBasis.bas"
end
To give an example of how this format looks like here is a part of the 3-21GSP basis of Buenker and
coworkers [295, 296]:
3-21GSP the name of the basis

! all these lines in the beginning with ! are comments
! BASIS="3-21GSP"
!Elements References
!-------- ----------
! H - Ne: A.V. Mitin, G. Hirsch, R. J. Buenker, Chem. Phys. Lett. 259, 151 (1996)
! Na - Ar: A.V. Mitin, G. Hirsch, R. J. Buenker, J. Comp. Chem. 18, 1200 (1997).
!
HYDROGEN ! (3s) -> [2s]
S 2
1 4.50036231 0.15631167
2 0.68128924 0.90466909
S 1
1 0.15137639 1.00000000
CARBON ! (6s,3p) -> [3s,2p]
S 3
1 499.24042249 0.03330322
2 75.25419194 0.23617745
3 16.86538669 0.81336259
L 2 L shells are a s and a p shell with identical exponents
1 0.89739483 0.24008573 0.46214684
2 0.21746772 0.81603757 0.66529098
L 1
1 4.52660451 1.00000000 1.00000000
STOP
In this way you can construct your favorite standard basis set and load it easily into the program. A word of
caution: in C/C++ the backslashes in directory assignments must be given twice to be correctly understood!
The file format for the auxiliary basis sets is exactly the same.
9.3.6 Advanced Specification of Effective Core Potentials
Library ECPs and Basis Sets
Besides the simple input line (section 6.3.3), assignment of ECPs and associated basis sets can be done within
the %basis block using the ECP and NewECP keywords.
In its most simple format, the NewECP keyword is used as in the following example:
%basis
ECP "def2-ECP" // All elements (for which the ECP is defined)
NewECP Pt "def2-SD" end
end
Similarly for a Los Alamos ECP and valence basis set:
NewECP Fe "LANL2" end
Note that the second argument in the above examples (the valence basis set) is actually optional in terms
of program functionality; only the ECP name is compulsory. However, no corresponding valence basis set
is assigned automatically by the program if none is specified. A library valence basis set can be explicitly
requested here, or alternatively any custom basis set can be provided manually with the NewGTO command.
A variant of the NewECP keyword can be used for individual atoms inside the geometry definition:
* xyz ...
...
S 0.0 0.0 0.0 NewECP "SDD" end
...
*
Manual Input of ECP Parameters
To manually specify ECP parameters, the NewECP keyword is followed by the element for which an ECP is
to be entered, the number of core electrons to be replaced (N core) and the maximum angular momentum
(lmax). The ECP specification is finished by giving the definitions of the individual shells that constitute the
angular dependent potentials Ul .
%basis
NewECP element
N_core (number of core electrons)
lmax (max. angular momentum)
[shells]
end
end
For each ECP shell, first the angular momentum l has to be given, followed by the number of primitives.
The primitives themselves are then specified by giving a running index and the respective tuple of exponent
akl , expansion coefficient dkl and radial power nkl .
# ECP shell
l (number of primitives)
1 a1l d1l n1l
2 a2l d2l n3l
...
As an example, consider the SD(10,MDF) for Vanadium. The name indicates a StuttgartDresden type ECP
that replaces 10 core electrons and is derived from a relativistic calculation for the neutral atom. It consists
of 4 shells with angular momentum s, p, d and f . Note that the f shell has an expansion coefficient of 0.0
and thus will not contribute at all to this effective core potential. This is typical for all SD potentials (but
may be different for program packages like TURBOMOLE that do not support arbitrary angular momentum
with respect to the ECP and therefore use recontractions of the original parameter sets).
# ECP SD(10,MDF) for V

# M. Dolg, U. Wedig, H. Stoll, H. Preuss,
# J. Chem. Phys. 86, 866 (1987).
NewECP V
N_core 10
lmax f
s 2
1 14.4900000000 178.4479710000 2
2 6.5240000000 19.8313750000 2
p 2
1 14.3000000000 109.5297630000 2
2 6.0210000000 12.5703100000 2
d 2
1 17.4800000000 -19.2196570000 2
2 5.7090000000 -0.6427750000 2
f 1
1 1.0000000000 0.0000000000 2
end
9.3.7 Embedding Potentials
Computations on cluster models sometimes require the presence of embedding potentials in order to account
for otherwise neglected repulsive terms at the border [297]. In order to simplify these kind of calculations
with ORCA the ECP embedding can be accomplished quite easy:
* ...
# atom> charge x y z optional ECP declaration
Zr> 4.0 0.0 0.0 0.0 NewECP "SDD" end
...
*
The declaration of such a coreless ECP centre takes place in the coordinates section by appending a bracket
> to the element symbol. Note that embedding ECPs are treated as point charges in ORCA, so the charge
has to be given next. The coordinates of the coreless ECP centre have to be specified as usual and may be
9.4 Choice of Initial Guess and Restart of SCF Calculations 339
followed by an optional ECP assignment. In general, calculations that employ an ECP embedding procedure
should be single point calculations. However if the need arises to perform a geometry optimization, make
sure to set up explicit Cartesian constraints for the coreless ECP centres.
9.4 Choice of Initial Guess and Restart of SCF Calculations
The initial guess is an important issue in each SCF calculation. If this guess is reasonable the convergence of
the procedure will be much better. ORCA makes some effort to provide a good initial guess and give the
user enough flexibility to tailor the initial guess to his or her needs.
The initial guess is also controlled via the %scf block and the variables Guess, MOInp and GuessMode.
%scf Guess HCore # One electron matrix

Hueckel # Extended H
uckel guess
PAtom # Polarized atomic densities
PModel # Model potential
MORead # Restart from an earlier calc.
MOInp "Name.gbw" # orbitals used for MORead
GuessMode FMatrix # FMatrix projection
CMatrix # Corresponding orbital projection
AutoStart true # try to use the orbitals from the existing
# GBW file of the same name (if possible)
# (default)
false # dont use orbitals from existing GBW file
end
9.4.1 AutoStart feature
Previous versions of ORCA always created a new GBW file at the beginning of the run no matter whether a
file of the same name existed or perhaps contained orbitals. This philosophy was changed in version 2.5-03.
The program now automatically checks if a .gbw file of the same name exists. If yes, the program checks
if it contains orbitals and all other necessary information for a restart. If yes, the variable Guess is set
to MORead. The existing .gbw file is renamed to BaseName.ges and MOInp is set to this filename. If the
AutoStart feature is not desired set AutoStart false in the %scf block or give the keyword !NoAutoStart
in the simple input line.
9.4.2 One Electron Matrix Guess
The simplest guess is to diagonalize the one electron matrix to obtain starting orbitals. This guess is very
simple but usually also a disaster because it produces orbitals that are far too compact.
9.4.3 Basis Set Projection
The remaining guesses (may) need the projection of initial guess orbitals onto the actual basis set. In ORCA
there are two ways this can be done. GuessMode FMatrix and GuessMode CMatrix The results from the
two methods are usually rather similar. In certain cases GuessMode CMatrix may be preferable. GuessMode
FMatrix is simpler and faster. In short the FMatrix projection defines an effective one electron operator:
i ai ai
X
f= (9.46)
i
where the sum is over all orbitals of the initial guess orbital set, ai is the creation operator for an electron
in guess MO i, ai is the corresponding destruction operator and i is the orbital energy. This effective one
electron operator is diagonalized in the actual basis and the eigenvectors are the initial guess orbitals in the
target basis. For most wavefunctions this produces a fairly reasonable guess.
CMatrix is more involved. It uses the theory of corresponding orbitals to fit each MO subspace (occupied,
partially occupied or spin-up and spin-down occupied) separately [298,299]. After fitting the occupied orbitals
the virtual starting orbitals are chosen in the orthogonal complement of the occupied orbitals. In some cases,
especially when restarting ROHF calculations this may be an advantage. Otherwise I have not met cases
where CMatrix was grossly superior to FMatrix but that does not mean that such cases do not exist and
they may not even be exotic.
9.4.4 PModel Guess
The PModel guess (chosen by Guess PModel in the %scf block or simply a keywordline with !PModel) is
one that is usually considerably successful. It consists of building and diagonalizing a Kohn-Sham matrix
with an electron density which consists of the superposition of spherical neutral atoms densities which are
predetermined for both relativistic and nonrelativistic methods. This guess is valid for both Hartree-Fock
and DFT methods but not for semiempirical models. However, due to the complexity of the guess it will also
take a little computer time (usually less than one SCF iteration). The model densities are available for most
atoms of the periodic table and consequently the PModel guess is usually the method of choice (particularly
for molecules containing heavy elements) unless you have more accurate starting orbitals available.
9.4.5 H
uckel and PAtom Guesses
The extended H uckel guess proceeds by performing a minimal basis extended H uckel calculation and projecting
the MOs from this calculation onto the actual basis set using one of the two methods described above. The
minimal basis is the STO-3G basis set. The H uckel guess may not be very good because the STO-3G basis set
is so poor. There is also accumulating evidence that the superposition of atomic densities produces a fairly
good initial guess. The critique of the atomic density method is that the actual shape of the molecule is not
taken into account and it is more difficult to reliably define singly occupied orbitals for ROHF calculations or
a reasonable spin density for UHF calculations. Therefore ORCA chooses a different way in the PAtom guess
(which is the default guess) - the H
uckel calculation is simply carried out for all electrons in a minimal basis
of atomic SCF orbitals. These were determined once and for all and are stored inside the program. This
9.4 Choice of Initial Guess and Restart of SCF Calculations 341
means that the densities around the atoms are very close to the atomic ones, all orbitals on one center are
exactly orthogonal, the initial electron distribution already reflects the molecular shape and there are well
defined singly occupied orbitals for ROHF calculations. This guess has to the best of my knowledge not been
described before.
9.4.6 Restarting SCF Calculations
To restart SCF calculations, it can be very helpful and time-saving to read in the orbital information of a
previous calculation. To do this, specify:
! moread
% moinp "name.gbw"
Starting from version 2.5-03 of ORCA this is done by default if the .gbw file of the same name exists.
The program stores the current orbitals in every SCF cycle. Should a job crash, it can be restarted from
the orbitals that were present at this time by just re-running the calculation to use the present .gbw file. In
addition, an effort has been made to make .gbw files from different releases compatible with each other. If
your input .gbw file is from an older release, use ! rescue moread noiter with % moinp "name.gbw" to
produce an up-to-date .gbw. When the rescue keyword is invoked, only the orbital coefficients are read from
the .gbw file, and everything else from the input file. Thus, make sure that the geometry and the basis set of
the old .gbw file and the new input match.
Within the same ORCA version, neither the geometry nor the basis set stored in name.gbw need to match the
present geometry or basis set. The program merely checks if the molecules found in the current calculation
and name.gbw are consistent with each other and then performs one of the possible orbital projections. If
the two basis sets are identical the program by default only reorthogonalizes and renormalizes the input
orbitals. However, this can be overruled by explicitly specifying GuessMode in the % scf block as CMatrix
or FMatrix.
For pre 2.5-03 versions of ORCA the input .gbw file from the earlier calculation must have a different name
than the new calculation, because in the very beginning of a calculation, a new .gbw file is written. If the
names are the same, the .gbw file from the earlier calculation is overwritten and all information is lost.
Therefore, if you want to restart a calculation with an input file of the same name as the previous calculation,
you have to rename the .gbw file first.
There is an additional aspect of restarting SCF calculations if you have chosen SCFMode = Conventional or
SCFMode = SemiDirect the program stores a large number of integrals that might have been time consuming
to calculate on disk. Normally the program deletes these integrals at the end of the calculation. However, if
you want to do a closely related calculation that requires the same integrals (i.e. the geometry, the basis
set and the threshold Thresh are the same) it is faster to use the integrals generated previously. This is
done by using KeepInts = true in the % scf block of the first calculation and then use ReadInts = true
in the the % scf block of the second calculation. If the second calculation has a different name than the first
calculation you have to use IntName = "FirstName" to tell the program the name of the integral files. Note
that the file containing the integrals does not have an extension it is simply the name of the previous
input file with .inp stripped off.
%scf KeepInts true # Keep integrals on disk

ReadInts true # Read integrals from disk
IntName "MyInts" # Name of the integral files without extension
end
Note that if you want to reuse the integrals for SemiDirect calculations it is forbidden to change the values
of TCost and TSize in the % scf block. The program will not check for that but your results will be totally
meaningless. In general, restarting calculations with old integral files requires the awareness and responsibility
of the user. If properly used, this feature can save considerable amounts of time.
9.4.7 Changing the Order of Initial Guess MOs and Breaking the Initial Guess
Symmetry
Occasionally you will want to change the order of initial guess MOs be it because the initial guess yielded
an erroneous occupation pattern or because you want to converge to a different electronic state using the
orbitals of a previous calculation. Reordering of MOs and other tasks (like breaking the symmetry of the
electronic wavefunction) are conveniently handled with the Rotate feature in ORCA. Rotate is a subblock
of the SCF block that allows you to linearly transform pairs of MOs.
%scf
Rotate
{ MO1, MO2, Angle }
{ MO1, MO2, Angle, Operator1, Operator2 }
end
end
Here, MO1 and MO2 are the indices of the two MOs of interest. Recall that ORCA starts counting MOs with
index 0, i.e. the MO with index 1 is the second MO. Angle is the rotation angle in degrees. A rotation
angle of 90 corresponds to flipping two MOs, an angle of 45 leads to a 50:50 mixture of two MOs, and a
180 rotation leads to a change of phase. Operator1 and Operator2 are the orbitals sets for the rotation.
For UHF calculations spin-up orbitals belong to operator 0 and spin-down orbitals to operator 1. RHF and
ROHF calculations only have a single orbital set.
The Rotate feature can be used to produce broken symmetry solutions say in transition metal dimers. In
order to do that first perform a high-spin calculation, then find the pairs of MOs that are symmetric and
antisymmetric combinations of each other. Take these MOs as the initial guess and use rotations of 45 for
each pair to localize the startup MOs. If you are lucky and the broken symmetry solution exists, you have a
good chance of finding it this way. The physical meaning of such a solution is a different question that will
not be touched here.
9.5 SCF Convergence 343
9.5 SCF Convergence
SCF convergence is a pressing problem in any electronic structure package because the total execution times
increases linearly with the number of iterations. Thus, it remains true that the best way to enhance the
performance of an SCF program is to make it converge better. In some cases, especially for open-shell
transition metal complexes, convergence may be very difficult. ORCA makes a dedicated effort to achieve
reasonable SCF convergence for these cases without compromising efficiency. However, there are cases that
do not converge, but find a list of things to try first before giving up in this chapter.
Another issue is whether the solution found by ORCA is stable, i.e. a minimum on the surface of orbital
rotations. Especially for open-shell singlets it can be hard to achieve a broken-symmetry solution. The SCF
stability analysis (section 9.7) may be able to help in such situations.
9.5.1 Convergence Tolerances
Before discussing how to converge a SCF calculation it should be defined what is meant by converged.
ORCA has a variety of options to control the target precision of the energy and the wavefunction that can be
selected in the % scf block, or with a simple input line keyword that merges the criterion label with SCF,
e.g. ! StrongSCF or ! VeryTightSCF:
%scf
Convergence # The default convergence is between medium and strong
Sloppy # very weak convergence
Loose # still weak convergence
Medium # intermediate accuracy
Strong # stronger
Tight # still stronger
VeryTight # even stronger
Extreme # close to numerical zero of the computer
# in double precision arithmetic
end
Like other keys, Convergence is a compound key that assigns default values to a variety of other variables
given in the box below. The default convergence criteria are reasonable and should be sufficient for most
purposes. For a cursory look at populations weaker convergence may be sufficient, whereas other cases may
require stronger than default convergence. Note that Convergence does not only affect the target convergence
tolerances but also the integral accuracy as discussed in the section about direct SCF and alike. This is very
important: if the error in the integrals is larger than the convergence criterion, a direct SCF calculation
cannot possibly converge.
The convergence criteria are always printed in the output. Given below is a list of the convergence criteria
for ! TightSCF, which is often used for calculations on transition metal complexes.
%scf
TolE 1e-8 # energy change between two cycles
TolRMSP 5e-9 # RMS density change
TolMaxP 1e-7 # maximum density change
TolErr 5e-7 # DIIS error convergence
TolG 1e-5 # orbital gradient convergence
TolX 1e-5 # orbital rotation angle convergence
ConvCheckMode 2 # = 0: check all convergence criteria
# = 1: stop if one of criterion is met, this is sloppy!
# = 2: check change in total energy and in one-electron energy
# Converged if delta(Etot)<TolE and delta(E1)<1e3*TolE
ConvForced # = 0: convergence not mandatory for next calculation step
# = 1: break, if you did not meet the convergence criteria
end
If ConvCheckMode=0, all convergence criteria have to be satisfied for the program to accept the calculation
as converged, which is a quite rigorous criterion. In this mode, the program also has mechanisms to decide
that a calculation is converged even if one convergence criterion is not fulfilled but the others are overachieved.
ConvCheckMode=1 means that one criterion is enough. This is quite dangerous, so ensure that none of the
criteria are too weak, otherwise the result will be unreliable. The default ConvCheckMode=2 is a check of
medium rigour the program checks for the change in total energy and for the change in the one-electron
energy. If the ratio of total energy and one-electron energy is constant, the self-consistent field does not
fluctuate anymore and the calculation can be considered converged. If you have small eigenvalues of the
overlap matrix, the density may not be converged to the number of significant figures requested by TolMaxP
and TolRMSP.
ConvForced is a flag to prevent time consuming calculations on non-converged wave functions. It will default
to ConvForced=1 for Post-HF methods, Excited States runs and Broken Symmetry calculations. You can
overwrite this default behavior by setting ConvForced=0.
Irrespective of the ConvForced value that has been chosen, properties or numerical calculations (NumGrad,
NumFreq) will not be performed on non-converged wavefunctions!
9.5.2 Dynamic and Static Damping
Damping is the oldest and simplest convergence aid. It was already invented by Douglas Hartree when he did
his famous atomic calculations. Damping consists of mixing the old density with the new density as:
Pnew, damped = (1 ) Pnew + Pold (9.47)
where is the damping factor, which must have a value of less than 1. Thus the permissible range (not
checked by the program) is 0 . . . 0.999999. For values larger than 1, the calculation cannot proceed since no
new density is admixed. Damping is important in the early stages of a calculation where Pold and Pnew are
very different from each other and the energy is strongly fluctuating. Many schemes have been suggested that
vary the damping factor dynamically to give strong damping in the beginning and no damping in the end of
an SCF. The scheme implemented in ORCA is that by Hehenberger and Zerner [300] and is invoked with
CNVZerner=true. Static damping is invoked with CNVDamp=true. These convergers are mutually exclusive.
They can be used in the beginning of a calculation when it is not within the convergence radius of DIIS or
SOSCF. Damping works reasonably well, but most other convergers in ORCA are more powerful.
If damping used in conjunction with DIIS or SOSCF, the value of DampErr is important: once the DIIS error
falls below DampErr, the damping is turned off. In case SOSCF is used, DampErr refers to the orbital gradient
value at which the damping is turned off. The default value is 0.1 Eh. In difficult cases, however, it is a good
idea to choose DampErr much smaller, e.g. 0.001. This is to some extent chosen automatically together
with the keyword ! SlowConv.
%scf
# control of the Damping procedure
CNVDamp true # default: true
CNVZerner false # default: false
DampFac 0.98 # default: 0.7
DampErr 0.05 # default: 0.1
DampMin 0.1 # default: 0.0
DampMax 0.99 # default: 0.98
# more convenient:
Damp fac 0.98 ErrOff 0.05 Min 0.1 Max 0.99 end
end
9.5.3 Level Shifting
Level shifting is a frequently used technique. The basic idea is to shift the energies of the virtual orbitals such
that after diagonalization the occupied and virtual orbitals mix less strongly and the calculation converges
more smoothly towards the desired state. Also, level shifting should prevent flipping of electronic states
in near-degenerate cases. In a special context it has been shown by Saunders and Hillier [301, 302] to be
equivalent to damping.
Similar to DampErr described in the previous section, ShiftErr refers to the DIIS error at which the level
shifting is turned off.
%scf
# control of the level shift procedure
CNVShift true # default: true
LShift 0.1 # default: 0.25, energy unit is Eh.
ShiftErr 0.1 # default: 0.0
# more convenient:
Shift Shift 0.1 ErrOff 0.1 end
end
9.5.4 Direct Inversion in Iterative Subspace (DIIS)
The Direct Inversion in Iterative Subspace (DIIS) is a technique that was invented by Pulay [303, 304]. It
has become the de facto standard in most modern electronic structure programs, because DIIS is robust,
efficient and easy to implement. Basically DIIS uses a criterion to judge how far a given trial density is from
self-consistency. The commutator of the Fock and density matrices [F,P] is a convenient measure for this
error. With this information, an extrapolated Fock matrix from the present and previous Fock matrices is
constructed, which should be much closer to self-consistency. In practice this is usually true, and better than
linear convergence has been observed with DIIS. In some rare (open-shell) cases however, DIIS convergence is
slow or absent after some initial progress. As self-consistency is approached, the set of linear equations to be
solved for DIIS approaches linear dependency and it is useful to bias DIIS in favor of the SCF cycle that
had the lowest energy using the factor DIISBfac. This is achieved by multiplying all diagonal elements of
the DIIS matrix with this factor unless it is the Fock matrix/density which leads to the lowest energy. The
default value for DIISBfac is 1.05.
The value of DIISMaxEq is the maximum number of old Fock matrices to remember. Values of 5-7 have been
recommended, while other users store 10-15 Fock matrices. Should the standard DIIS not achieve convergence,
some experimentation with this parameter can be worthwhile. In cases where DIIS causes problems in the
beginning of the SCF, it may have to be invoked at a later stage. The start of the DIIS procedure is controlled
by DIISStart. It has a default value of 0.2 Eh, which usually starts DIIS after 0-3 cycles. A different way of
controlling the DIIS start is adjusting the value DIISMaxIt, which sets the maximum number of cycles after
which DIIS will be started irrespective of the error value.
%scf
# control of the DIIS procedure
CNVDIIS true # default: true
DIISStart 0.1 # default: 0.2
DIISMaxIt 5 # default: 12
DIISMaxEq 7 # default: 5
DIISBFac 1.2 # default: 1.05
DIISMaxC 15.0 # default: 10.0
# more convenient:
DIIS Start 0.1 MaxIt 5 MaxEq 7 BFac 1.2 MaxC 15.0 end
end
9.5.5 An alternative DIIS algorithm: KDIIS
An alternative algorithm that makes use of the DIIS concept is called KDIIS in ORCA. The KDIIS algorithm
is designed to bring the orbital gradient of any energy expression to zero using a combination of DIIS
extrapolation and first order perturbation theory. Thus, the method is diagonalization-free. In our hands it
is superior to the standard DIIS algorithm in many cases, but not always. The algorithm is invoked with the
keyword ! KDIIS and is available for RHF, UHF and CASSCF.
9.5.6 Approximate Second Order SCF (SOSCF)
SOSCF is an approximately quadratically convergent variant of the SCF procedure [305, 306]. The theory is
relatively involved and will not be described here. In short SOSCF computes an initial guess to the inverse
orbital hessian and then uses the BFGS formula in a recursive way to update orbital rotation angles. As
information from a few iterations accumulates, the guess to the inverse orbital hessian becomes better and
better and the calculation reaches a regime where it converges superlinearly. As implemented, the procedure
converges as well or slightly better than DIIS and takes a somewhat less time. However, it is also a lot
less robust, so that DIIS is the method of choice for many problems (see also the description of the full
Newton-Raphson procedure in the next section). On the other hand, SOSCF is useful when DIIS gets stuck
at some error around 0.001 or 0.0001. Such cases were the primary motive for the implementation of
SOSCF into ORCA.
The drawback of SOSCF is the following: in the beginning of the SCF, the orbital gradient (the derivative of
the total energy with respect to rotations that describe the mixing of occupied and virtual MOs) is large,
so that one is far from the quadratic regime. In such cases, the procedure is not successful and may even
wildly diverge. Therefore it is recommended to only invoke the SOSCF procedure in the very end of the
SCF where DIIS may lead to trailing convergence. SOSCF is controlled by the variables SOSCFStart and
SOSCFMaxIt. SOSCFStart is a threshold for the orbital gradient. When the orbital gradient, or equivalently
the DIIS Error, fall below SOSCFStart, the SOSCF procedure is initiated. SOSCFMaxIt is the latest iteration
to start the SOSCF even if the orbital gradient is still above SOSCFStart.
%scf
# control of the SOSCF procedure
CNVSOSCF true # default: false
SOSCFStart 0.1 # default: 0.01
SOSCFMaxIt 5 # default: 1000
# more convenient:
SOSCFStart 0.1 MaxIt 5 end
end
For many calculations on transition metal complexes, it is a good idea to be conservative in the startup
criterion for SOSCF, it may diverge otherwise. A choice of 0.01 or lower can be recommended.
9.5.7 Full Second Order SCF (Newton-Raphson SCF)
The NewtonRaphson (NR) procedure is quite powerful once the calculations is within its radius of convergence:
it then usually converges extremely fast and to high accuracy. It should be considered that the time
requirement for a NR iteration is much higher than that for a regular SCF calculation. However, this may
be overcompensated by the reduced number of cycles (see section 8.1.5 for a numerical example). For a
normal well-behaved molecule which converges in 1015 iterations with the DIIS procedure, it is unlikely
that invoking the NR converger pays off. For more problematic cases it is more likely that it will. Another
area where NR-SCF may be useful is when you desire accurate SCF solutions, for example for numerical
frequency calculations. In these cases, a different option is to decrease the convergence tolerance of the
CP-SCF solver (Z Tol).
In version 2.4.45 the full NewtonRaphson method (and its augmented Hessian analogue) was implemented
for converging closed-shell or spin-unrestricted HartreeFock and density functional calculations.
The NR solver itself has two options, which are very similar to the analogous SOSCF options:
%scf
# no Newton-Raphson by default
NRMaxIt 10 # iteration at which NR is switched on, default 150
NRStart 0.05 # DIIS error at which NR is switched on, default 0.0033
end
end
As soon as the NewtonRaphson procedure starts within one SCF procedure, all other convergers are switched
off. After convergence is achieved, the orbitals are canonicalized again through a single diagonalization of the
Fock operator. In order to help the convergence of the CP-SCF procedure, the orbitals are quasi-canonical in
between iterations. This means that after the orbital update, the occocc and virtvirt blocks of the Fock
operator are diagonalized separately and the occupied and virtual orbitals are chosen to be eigenfunctions of
these parts of the Fock operator.
There is a shortcut to the NR-converger. If ! NRSCF is chosen, this will invoke the NRSCF once the DIIS
error falls below 0.02 (the default value). If SlowConv is also specified, the NR converger will only be switched
on at a DIIS error of 0.002. This is more conservative since the NR-solver may also have convergence problems
otherwise.
In particularly difficult cases, the NewtonRaphson step may run into trouble if the orbital Hessian is singular
or nearly singular. It is then preferable to take a more cautious step, by invoking the augmented-Hessian
method which was implemented for such cases. This method is called by ! AHSCF. It sets the same values as
NRSCF but puts an augmented Hessian solver into operation. It will not always help, but in some difficult
situations this feature has already been successful.
Technically speaking, the program solves a large set of linear equations in each NR iteration. It turns out
that the equations to be solved are precisely equivalent to the CP-SCF equations for an electric-field-like
equation but with the right hand side being replaced by the occupied/virtual blocks of the Fock operator in
the MO basis. Thus, the implementation calls the CP-SCF program in each SCF iteration. The converged
CP-SCF first order wavefunction coefficients are then used to update the current orbitals. Thus, the cost of
each NR iteration can be precisely specified as the cost of one electric field type CP-SCF calculation for a
single perturbation on the same system.
One important aspect of the NR implementation is that the NR solver checks itself for convergence. The
orbital gradient is calculated before the CP-SCF program is called. If the orbital gradient is below its
threshold (TolG), convergence is signaled irrespective of what the other convergence indicators are. The
orbital gradient is a rigorous criterion and therefore the results are considered reliable. If tighter convergence
is desired, the thresholds TolG and TolX should be lowered.
The following example for a normal case of a HartreeFock geometry optimization converges in 5 cycles:
! RHF TZVPP Opt VeryTightSCF
* xyz 0 1
N 0 0 0
N 0 0 1
*
With the alternative first order SCF it also takes 5 cycles, but in each geometry optimization step there
are more SCF iterations. In these examples, the total time for the calculations is comparable (less than 30
seconds), but that may be different for other cases.
! RHF TZVPP Opt TightSCF

! NRSCF
* xyz 0 1
N 0 0 0
N 0 0 1
*
The CP-SCF program is called with the convergence and method flags that are also used in Z-vector
calculations (e.g. in MP2 or CIS/TD-DFT gradient calculations). They are accessible through the %method
block:
%method Z_solver Pople # Poples solver. Usually great!

CG # conjugate gradient
DIIS # Pulays DIIS
AugHess # Augmented Hessian (for SCF only)
Z_MaxIter 64 # maximum number of CP-SCF iterations
Z_MaxDIIS 10 # for DIIS: maximum number of expansion vectors
Z_Shift 0.3 # for DIIS: level shift
Z_Tol 1e-6 # convergence tolerance on residual
end
The range of applicability of the NR method is precisely the same as that of the CP-SCF program itself and
the SCF program will immediately benefit from any improvement made in the CP-SCF module.
9.5.8 Finite Temperature HF/KS-DFT
A finite temperature can be used to apply a Fermi-like occupation number smearing over the orbitals of
the system, which may sometimes help to get convergence of the SCF equations in near hopeless cases.
Through the smearing, the electrons are distributed according to Fermi statistics among the available orbitals.
The chemical potential is found through the condition that the total number of electrons remains correct.
Gradients can be computed in the presence of occupation number smearing.
%scf SmearTemp 5000 # temperature in Kelvin

end
NOTE:
Finite temperature SCF (fractional occupation numbers or FOD analysis, see 9.5.8.1 and 9.5.8.2,
respectively) cannot be used together with the CNVRico or SOSCF methods.
9.5.8.1 Fractional Occupation Numbers
Only a very basic implementation of fractional occupation numbers is presently provided. It is meant to deal
with orbitally degenerate states in the UHF/UKS method. Mainly it was implemented to avoid symmetry
breaking in DFT calculations on orbitally degenerate molecules and atoms. The program checks the orbital
energies of the initial guess orbitals, finds degenerate sets and averages the occupation numbers among them.
Currently the criterion for degenerate orbitals is 103 Eh. The fractional occupation number option is invoked
by:
%scf FracOcc true

end
Clearly, the power of fractional occupation numbers goes far beyond what is presently implemented in the
program and future releases will likely make more use of them. The program prints a warning whenever it
uses fractional occupation numbers. The fractionally occupied orbitals should be checked to ensure they are
actually the intended ones.
NOTE:
Using GuessMode = CMatrix will cause problems because there are no orbital energies for the initial
guess orbitals. The program will then average over all orbitals which makes no sense at all.
9.5.8.2 Fractional Occupation Number Weighted Electron Density (FOD)
Many approximate QC methods do not yield reliable results for systems with significant static electron
correlation (SEC) but it is often difficult to predict if the system in question suffers from SEC or not. Existing
scalar SEC diagnostics (e.g., the T1 diagnostic) do not provide any information where the SEC is located in the
molecule. Furthermore, often quite expensive calculations have to be performed first (e.g., CCSD) in order to
judge the reliability of the results based on a single number. Molecular systems with strong SEC (e.g. covalent
bond-breaking, biradicals, open-shell transition metal complexes) are usually characterized by small energy
gaps between frontier orbitals, and hence, the appearance of many equally important determinants in their
electronic wavefunction. This finding is used in the FOD analysis [307] which is based on finite temperature
KS-DFT where the fractional occupation numbers are determined from the Fermi distribution (Fermi
smearing)
1
fi =
e(i EF )/kTel +1
The central quantity of the FOD analysis is the fractional occupation number weighted electron density
(F OD ), a real-space function of the position vector r:
N
X
F OD (r) = (1 2 fi )|i (r)|2
i
(1 and 2 are unity if the level is lower than EF while they are 0 and 1, respectively, for levels higher than
EF ). The fi represent the fractional occupation numbers (0 fi 1; sum over all electronic single-particle
levels obtained by solving self-consistently the KS-SCF equations minimizing the free-electronic energy).
F OD (r) can be plotted using a pre-defined contour surface value (see 9.31.2.2). FOD plots only show the
contribution of the hot (strongly correlated) electrons and can thus be used to choose a suitable QC method
for the system in question based on some rules of thumb (see 9.31.2.2). Mulliken reduced orbital charges
based on F OD (r) (see 9.30.2) offer a fast alternative to get the informations of the FOD plot.
The integration of F OD over all space yields as additional information a single size-extensive number termed
NF OD which correlates well with other scalar SEC diagnostics and can be used to globally quantify SEC
effects in the molecule.
F OD (and NF OD ) strongly depend on the orbital energy gap which itself depends almost linearly on the
amount of the non-local Fock exchange admixture ax . The following (empirical) function of the optimal
electronic temperature Tel on ax
Tel = 20000 K ax + 5000 K
is used to ensure that similar results of the FOD analysis are obtained with various functionals. For example,
the SmearTemp has to be 5000 K for TPSS (ax = 0), 9000 K for B3LYP (ax = 20%), 10000 K for PBE0 (ax
= 25%), and 15800 K for M06-2x (ax = 54%). The result of the FOD analysis is not strongly dependent
on the employed basis set (see supplementary information of the original publication [307]). TPSS/def2-
TZVP/TightSCF was chosen as the default since it is fast and robust. The FOD analysis is a very efficient
and practicable tool to get informations about the amount and localization of SEC in the system of question.
It is called by a simple keyword:
# ground state of p-benzyne

! FOD
* xyz 0 1
C 0.0000000 1.2077612 0.7161013
C 0.0000000 0.0000000 1.3596219
C 0.0000000 -1.2077612 0.7161013
C 0.0000000 -1.2077612 -0.7161013
C 0.0000000 0.0000000 -1.3596219
C 0.0000000 1.2077612 -0.7161013

H 0.0000000 2.1606260 1.2276695
H 0.0000000 -2.1606260 1.2276695
H 0.0000000 -2.1606260 -1.2276695
H 0.0000000 2.1606260 -1.2276695
*
The respective output reads:
--------------
SCF ITERATIONS
--------------
*** Starting incremental Fock matrix formation ***
0 -230.8802282249 0.000000000000 0.02185974 0.00097019 0.1125725 0.7000
1 -230.9307557407 -0.050527515790 0.02830189 0.00098494 0.0564588 0.7000
... etc.
12 -231.0034669835 0.000000000547 0.00000176 0.00000004 0.0000032 0.0000
***DIIS convergence achieved***
Fermi smearing: E(HOMO(Eh)) = -0.201006 MUE = -0.179109 gap = 1.117 eV
N_FOD = 0.921577
If the FOD analysis should be done employing a different functional, one has to explicitly specify the functional
and basis set in the simple keyword line and adjust the SmearTemp accordingly.
# ozone molecule
! UKS B3LYP def2-TZVP TightSCF
%scf
SmearTemp 9000
end
* xyz 0 1
O 0.00000000017911 0.00000000000000 0.43702029765776
O -1.09512651993192 0.00000000000000 -0.21851064888325
O 1.09512651975281 0.00000000000000 -0.21851064877451
*
The FOD analysis may also be useful for finding a suitable active space for e.g. CASSCF calculations.
NOTE:
The FOD analysis will be always printed (including Mulliken reduced orbital charges based on F OD )
if SmearTemp > 0 K F OD is stored on disk in the file Basename.scfp fod).
9.6 Choice of Wavefunction and Integral Handling 353
Since the the S2 expectation value is not defined for fractional occupation numbers, its printout is
omitted.
9.6 Choice of Wavefunction and Integral Handling
9.6.1 Choice of Wavefunction Type
The basic variable that controls the type of wavefunction to be computed is the variable HFTyp in the %scf
block. If nothing is specified for HFTyp, the program will check the multiplicity given in the input: for
closed-shell molecules with multiplicity 1, RHF/RKS is assumed; for open shell molecules with multiplicity
larger than 1, UHF/UKS is invoked. RHF will lead to a spin restricted closed-shell type computation [59].
For DFT calculations, RKS, UKS and ROKS can be used as synonyms for RHF, UHF and ROHF. The restricted
open-shell DFT method (ROKS) is only operative for high-spin states that have n unpaired electrons and
S = n/2. UKS wavefunctions will not be spin-purified.
%scf HFTyp RHF # closed-shell (RKS for DFT)

UHF # unrestricted open-shell (UKS for DFT)
ROHF # restricted open-shell (ROKS for DFT)
CASSCF # complete active space SCF
end
In certain cases you may want to run open-shell molecules with RHF/RKS to get a half-electron type
wavefunction [308]. The total energy is not corrected! Sometimes these half-electron computations lead to
acceptable convergence, and the resulting orbitals may be used as input for ROHF, UHF or MRCI calculations.
Especially for transition metal complexes the orbitals are quite different from ROHF or UHF orbitals, so that
it is not recommended to over-interpret the wavefunctions from such calculations. The calculation is set up
in the following way:
%method AllowRHF true end

# or simply: ! AllowRHF
For ROHF calculations [309317] the program will try to figure out what type of open-shell situation is
present on the basis of the initial guess orbitals and their energies. Most simple cases are well recognized
but sometimes a little help from the user is needed. The ROHF code has a very powerful feature that goes
back to insights of Mike Zerner [318, 319]. It can average over either all states of a given configuration (CAHF)
or all states of a given spin for a given configuration (SAHF). Especially the SAHF feature gives you easy access
to most degenerate high symmetry situations and the orbitals resulting from such calculations will be very
convenient as input for CI calculations. In this way one can approach results from MCSCF calculations.
The input variables described above are accessed in the %scf block. ROHFOPT Case User is a convenient way
to input details. For example for the high spin case with three electrons in three orbitals gives two operators
with vector coupling coefficients a = 1 and b = 2 (Zerner convention).
%scf HFTyp ROHF

ROHF_case CAHF # configuration averaged HF
SAHF # spin averaged HF
HighSpin # high spin case, n unpaired e- in n orbitals with S=n/2
ROHF_NumOp 3 # number of operators (3, 2 or 1)
ROHF_NOrb[1] 2,1 # number of orbitals in each open-shell
ROHF_NEl[1] 1,1 # number of electrons in each open-shell
end
ROHFOP Case User # manual input of ROHF variables
Nop 2 # number of operators
Norb[1] 3 # number of open-shell orbitals
Nel[1] 3 # number of open-shell electrons
A[1,1] 1 # Coulomb vector in the open shell
B[1,1] 2 # Exchange vector in the open shell
end
end
The hypothetical example below could represent an excited state of an octahedral d3 transition metal complex.
In this case there are five open-shell orbitals. The first three open-shell orbitals contain two electrons and the
last two one electron. The input for a SAHF calculation is identical, just replace CAHF with SAHF.
%scf HFTyp ROHF

ROHF_case CAHF # configuration averaged HF
ROHF_NumOp 3 # 3 operators in this case: closed, open1, open2
ROHF_NOrb[1] 3,2 # 3 orbitals in first open shell, 2 in the second
ROHF_NEl[1] 2,1 # 2 electrons in first open shell, 1 in the second
end
For a ROHF case HighSpin calculation an analogous input can be used, but the number of operators is always
two (one closed and one open-shell), therefore:
%scf HFTyp ROHF

ROHF_case HighSpin # high spin case
ROHF_NEl[1] 4 # n=4 unpaired electrons, S=n/2
end
One awkward feature of the ROHF theory is that the Fock operator is somewhat arbitrarily defined. Different
choices lead to the same wavefunction, but have different convergence properties that may vary from system
to system. ORCA thus lets the user choose the desired variant. Playing around with these choices may turn
a divergent or slowly converging ROHF calculation into a successful calculation!
The ROHF Restrict feature is another feature that may be useful. If you suspect that the ROHF calculation
does not converge because an open-shell and a closed-shell orbital are flipping back and forth, you can try to
avoid this behavior by choosing ROHF Restrict= true. Of course there is no guarantee that it will work,
and no guarantee that the system stays in the desired state. However, it decreases the chances of large,
uncontrolled steps.
%scf ROHF_Mode 0 # construct F according to Pulay (default)

1 # construct F as in the Gamess program
2 # construct F according to Kollmar
ROHF_Restrict false # restrict orbital interchanges and off-diagonal elements
# (default=false)
# a complete list of ROHF variables
ROHFOP Case User # manual input of ROHF variables
Nop 2 # number of operators
Norb[1] 3 # number of open-shell orbitals
Nel[1] 3 # number of open-shell electrons
A[1,1] 1 # Coulomb vector in the open shell
B[1,1] 2 # Exchange vector in the open shell
Mode 2 # use the Kollmar operator
Restrict false # do not restrict
end
end
9.6.2 UHF Natural Orbitals
The program can produce the UHF natural orbitals (UNOs). With these, the open-shell wavefunction can be
pictured conveniently. The syntax is simple:
%scf UHFNO true

end
# or simply: ! UNO
There are various printing options for UNOs described in the output section 9.30. The UNOs can also be
plotted as described in the plots section 9.31. In general the program stores a file BaseName.uno, where
BaseName is by default the name of you input file with .inp stripped off. Accordingly, the gbw file is named
BaseName.gbw. The .uno file is a normal gbw file that contains the geometry, basis set and the UNO orbitals.
It could be used, for example, to start a ROHF calculation.
9.6.3 Integral Handling (Conventional, Direct, SemiDirect)
As the number of nonzero integrals grows very rapidly and reaches easily hundreds of millions even with medium
sized basis sets in medium sized molecules, storage of all integrals is not generally feasible. This desperate
situation prevented SCF calculations on larger molecules for quite some time, so that Almlof [320322]
made the insightful suggestion to repeat the integral calculation, which was already the dominant step, in
every SCF cycle to solve the storage problem. Naively, one would think that this raises the effort for the
calculation to niter tintegrals (where niter is the number of iterations and tintegrals is the time needed to generate
the nonzero integrals). However, this is not the case because only the change in the Fock matrix is required
from one iteration to the next, but not the Fock matrix itself. As the calculations starts to converge, more
and more integrals can be skipped. The integral calculation time will still dominate the calculation quite
strongly, so that ways to reduce this burden are clearly called for. An important contribution to this subject
was made by Haser and Ahlrichs [323] who suggested to store some of the integrals and to only recalculate
the remaining ones in the direct SCF calculation. Obvious candidates for storage are those integrals that are
expensive to calculate. As integrals are calculated in batches 12 the cost of evaluating the given batch of shells
p, q, r, s may be estimated as:
cost np nq nr ns (2lp + 1) (2lq + 1) (2lr + 1) (2ls + 1) (9.48)
Here, np is the number of primitives involved in shell p, and lp is the angular momentum for this shell.
Large integrals are also good candidates for storage, because small changes in the density that multiply large
integrals are likely to give a nonzero contribution to the changes in the Fock matrix.
ORCA thus features three possibilities for integral handling, which are controlled by the variable SCFMode.
In the mode Conventional, all integrals above a given threshold are stored on disk (in a compressed format
that saves much disk space). In the mode Direct, all two-electron integrals are recomputed in each iteration.
In the mode SemiDirect, the H aser/Ahlrichs-type hybrid method as described above is implemented.
Two further variables are of importance: In the Conventional and SemiDirect modes the program may
write enormous amounts of data to disk. To ensure this stays within bounds, the program first performs a
so-called statistics run that gives a pessimistic estimate of how large the integral files will be. Oftentimes,
the program will overestimate the amount of disk space required by a factor of two or more. The maximum
amount of disk space that is allowed for the integral files is given by MaxDisk (in Megabytes).
On the other hand, if the integral files in Conventional and SemiDirect runs are small enough to fit into
the central memory, it is faster to do this since it avoids I/O bottlenecks. The maximum amount of memory
allocated for integrals in this way is specified by MaxIntMem (in Megabytes). If the integral files are larger
than MaxIntMem, no integrals will be read into memory.
%scf
MaxIter 100 # Max. no. of SCF iterations
SCFMode Direct # default, other choices: Conventional, SemiDirect
Thresh 1e-8 # Threshold for neglecting integrals / Fock matrix contributions
# Depends on the chosen convergence tolerance (in Eh).
TCut 1e-10 # Threshold for neglecting primitive batches. If the prefactor
# in the integral is smaller than TCut, the contribution of the
# primitive batch to the total batch is neglected.
TCost 100 # Threshold for considering integrals as costly in
# SemiDirect SCF. Batches with Cost > TCost are stored.
TSize 0.01 # In SemiDirect SCF: stores integrals with estim. size = TSize
12
A batch is a set of integrals that arises from all components of the shells involved in the integral. For example a
hpp|ppi batch gives rise to 3 3 3 3 = 81 integrals due to all possible combinations of px , py and pz functions
in the four shells. Computations based on batches lead to great computational advantages because the 81 integrals
involved in the hpp|ppi batch share many common intermediate quantities.
UseCheapInts false # default: false

DirectResetFreq 20 # default: 15
MaxDisk 2500 # Max. amount of disk for 2 el. ints. (MB)
MaxIntMem 400 # Max. amount of RAM for 2 el. ints. (MB)
end
The flag UseCheapInts has the following meaning: In a Direct or SemiDirect SCF calculation, the
oscillations in the total energy and density are initially quite large. High accuracy in the integrals is therefore
not crucial. If UseCheapInts is switched on, the program loosens the threshold for the integrals and thus
saves a lot of computational time. After having obtained a reasonable initial convergence, the thresholds are
tightened to the target accuracy. One pitfall with this method is that the number of cycles required to reach
convergence may be larger relative to a calculation with full integral accuracy throughout.13 When restarting
calculations that are close to convergence, it is recommended to switch UseCheapInts off. UseCheapInts
has no meaning in a conventional SCF.
The value of DirectResetFreq sets the number of incremental Fock matrix builds after which the program
should perform a full Fock matrix build in a Direct or SemiDirect SCF calculation. To prevent numerical
instabilities that arise from accumulated errors in the recursively build Fock matrix, the value should not be
too large, since this will adversely affect the SCF convergence. If the value is too small, the program will
update more frequently, but the calculation will take considerably longer, since a full Fock matrix build is
more expensive than a recursive one.
The thresholds TCut and Thresh also deserve a closer explanation. Thresh is a threshold that determines
when to neglect two-electron integrals. If a given integral is smaller than Thresh Eh, it will not be stored
or used in Fock matrix construction. Additionally, contributions to the Fock matrix that are smaller than
Thresh Eh will not be calculated in a Direct or SemiDirect SCF.
Clearly, it would be wasteful to calculate an integral, then find out it is good for nothing and thus discard it.
A useful feature would be an efficient way to estimate the size of the integral before it is even calculated, or
even have an estimate that is a rigorous upper bound on the value of the integral. Haser and Ahlrichs [323]
were the first to recognize that such an upper bound is actually rather easy to calculate. They showed that:
p p
|hij |kl i| 6 hij |ij i hkl |kl i (9.49)
where:
Z Z
1
hij |kl i = i (~r1 ) j (~r1 ) r12 k (~r2 ) l (~r2 ) d~r1 d~r2 (9.50)
Thus, in order to compute an upper bound for the integral only the right hand side of this equation must be
known. This involves only two index quantities, namely the matrix of two center exchange integrals hij |ij i.
These integrals are easy and quick to calculate and they are all >0 so that there is no trouble with the square
root. Thus, one has a powerful device to avoid computation of small integrals. In an actual calculation, the
Schwartz prescreening is not used on the level of individual basis functions but on the level of shell batches
13
This might be an undesirable feature of the current implementation.
because integrals are always calculated in batches. To realize this, the largest exchange integral of a given
exchange integral block is looked for and its square root is stored in the so called pre-screening matrix K
(that is stored on disk in ORCA). In a direct or semidirect SCF this matrix is not recalculated in every
cycle, but simply read from disk whenever it is needed. The matrix of exchange integrals on the level of
individual basis function is used in conventional or semidirect calculations to estimate the disk requirements
(the statistics run).
Once it has been determined that a given integral batch survives it may be calculated as:
X X X X
hij |kl i = dpi dqj dkr dsl hip jq |kr ls i (9.51)
p q r s
where the sums p, q, r, s run over the primitive gaussians in each basis function i, j, k, l and the ds are
the contraction coefficients. There are more powerful algorithms than this one and they are also used in
ORCA. However, if many terms in the sum can be skipped and the total angular momentum is low, it is still
worthwhile to compute contracted integrals in this straightforward way. In equation 9.51, each primitive
integral batch hip jq |kr ls i contains a prefactor Ipqrs that depends on the position of the four gaussians and
their orbital exponents. Since a contracted gaussian usually has orbital exponents over a rather wide range, it
is clear that many of these primitive integral batches will contribute negligibly to the final integral values. In
order to reduce the overhead, the parameter TCut is introduced. If the common prefactor Ipqrs is smaller than
TCut, the primitive integral batch is skipped. However, Ipqrs is not a rigorous upper bound to the true value
of the primitive integral. Thus, one has to be more conservative with TCut than with Thresh. In practice
it appears that choosing TCut=0.01*Thresh provides sufficient accuracy, but the user is encouraged to
determine the influence of TCut if it is suspected that the accuracy reached in the integrals is not sufficient.
HINT:
If the direct SCF calculation is close to convergence but fails to finally converge, this maybe related to
a numerical problem with the Fock matrix update procedure the accumulated numerical noise from
the update procedure prevents sharp convergence. In this case, set Thresh and TCut lower and/or let
the calculation more frequently reset the Fock matrix (DirectResetFreq).
NOTE:
For a Direct or SemiDirect calculation, there is no way to have Thresh larger than TolE. If the errors
in the Fock matrix are larger than the requested convergence of the energy, the change in energy can
never reach TolE. The program checks for that.
In a SemiDirect SCF, the number of stored integrals increases with molecular size. It may be necessary
to increase TSize and TCost to make storage possible. It is usually not a good idea to produce integral
files many gigabytes in size, because this will lead to quite some I/O penalty, especially in the late
stages of a calculation. Within reasonable bounds however, SemiDirect calculations can be quite a bit
faster than Direct calculations.
The actual disk space used for all temporary files may easily be larger than MaxDisk. MaxDisk only
pertains to the two-electron integral files. Other disk requirements are not currently checked by the
program and appear to be uncritical.
9.7 SCF Stability Analysis 359
9.7 SCF Stability Analysis
The SCF stability analysis evaluates the electronic Hessian (with respect to orbital rotations) at the point
indicated by the SCF solution to determine the lowest eigenvalues of the Hessian. If one or more negative
eigenvalues are found, the SCF solution corresponds to a saddle point and not a true local minimum in the
space considered in the analysis. A typical case are stretched bonds of diatomics, where the symmetry of the
initial guess leads to a restricted solution instead of the often preferred unrestricted one. Several spaces are
theoretically possible [139], however, ORCA limits itself to the analysis RHF/RKS in the space of UHF/UKS
or UHF/UKS in the space of UHF/UKS. As such, it is available for the SCF parts of DFT and HF. [140]
In the following, HF is used to indicate both HF and KS. Consider the following input (unless indicated
otherwise, default values are shown):
! BHLYP def2-SVP NORI
%scf
guess hcore # for illustrative purposes only
HFTyp UHF # default based on spin multiplicity
STABPerform true # default false
STABRestartUHFifUnstable true # restart the UHF-SCF if unstable
STABNRoots 3 # number of eigenpairs sought
STABMaxDim 3 # Davidson expansion space = MaxDim * NRoots
STABMaxIter 100 # maximum number of Davidson iterations
STABNGuess 4096 # size of initial guess matrix: 4096 x 4096
STABDTol 0.0001 # convergence criterion from iteration to iteration
STABRTol 0.0001 # convergence criterion max residual norm
STABlambda +0.5 # mixing parameter
STABORBWIN -1, -1, -1, -1, -1, -1, -1, -1 # four parameters in case of RHF
# orbital window, -1 refers to automatic determination
STABEWIN -5.0, 5.0 # lower and upper cutoff in Eh for automatic freezing
#---------------------------------------------------------------------------------
# alternative specification using a sub-block:
stab
NRoots 3
MaxDim 3 # etc.
end
end
* xyz 0 1
h 0.0 0.0 0.0
h 0.0 0.0 1.4
*
The determination of the electronic Hessian is structurally comparabale to the TDHF/CIS/TDDFT procedure.
Thus, many options are very similar and the user is encouraged to read the section on TDDFT (section 9.18)
to clarify some of the options given here. Since one is usually only interested in the qualitative determination
stable or not?, three roots should be sufficient to find the lowest eigenvalue. By the same philosophy,
MaxDim, MaxIter, NGuess and the convergence criteria were chosen. lambda refers to the of equation 37
of reference 139, which determines the mixing of the original SCF solution and the new orbitals to yield a
new guess. Choosing this value is not trivial, since positive and negative values can lead to different new
solutions (at least in principle). The convergence of the ensuing SCF depends on it, as well, since all SCF
procedures require a sufficiently good guess to converge in a decent number of iterations (or even at all).
The orbital window and the energy window can be specified. Note that the EWIN will be overridden by the
appropriate ORBWIN values. The automatic determination is also influenced by the %method FrozenCore
settings. Tests have shown that significant curtailing of the actual orbital window can drastically influence
the results to the point of qualitative failure.
Current limitations on the method are:
Only single-point-like calculations are supported. For geometry optimizations etc., one must use the
guess MORead feature 9.4 to employ the guess obtained here. Likewise, one must extract a geometry
and run a separate calculation if one is interested in the SCF stability.
As for TDDFT, NORI, RIJONX, and RIJCOSX are supported. RI-JK is not supported.
No solvation models are currently implemented to work with the stability analysis.
Other, more advanced features like finite-temperature calculations and relativistic calculations (beside
ECPs) are not possible at this time.
Overall, the user is cautioned against using the stability analysis blindly without critically evaluating the
result in terms of energy difference and by investigating the orbitals (by the printout or by plotting). Its
usefulness cannot be denied, but it is certainly not black-box.
9.8 Frozen Core Options
The frozencore (FC) approximation is usually applied in correlated calculation and consists in neglecting
correlation effects for electrons in the low-lying core orbitals. The FC approximation and the number of
core electrons per element can be adjusted in the %method block. The default number of core electrons per
element is listed in Table 9.8.
For systems containing heavy elements, core electrons might have higher orbital energies compared to the
orbital energies of valence MOs of some lighter elements. In that case, core electrons might be included in
the correlation calculation, which ultimately leads to large errors in correlation energy. In order to prevent
this, the MO ordering is checked after the SCF calculation: Do all lower energy MOs in the core region have
core electron character, i.e. are they strongly localized on the individual elements? If core orbitals are found
in the valence region, while more delocalized orbitals are found in the core region, the corresponding MO
pairs are swapped. The check for a correct frozen core can be switched off in the %method block.
9.8 Frozen Core Options 361
Table 9.8: Default values for number of frozen core electrons.
H He
0 0
Li Be B C N O F Ne
0 0 2 2 2 2 2 2
Na Mg Al Si P S Cl Ar
2 2 10 10 10 10 10 10
K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Br Kr
10 10 10 10 10 10 10 10 10 10 10 10 18 18 18 18 18 18
Rb Sr Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Te I Xe
18 18 28 28 28 28 28 28 28 28 28 28 36 36 36 36 36 36
Cs Ba Lu Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Po At Rn
36 36 46 46 46 46 46 46 46 46 46 46 68 68 68 68 68 68
Fr Ra Lr Rf Db Sg Bh Hs Mt Ds Rg Cn
68 68 68 100 100 100 100 100 100 100 100 100
Lanthanides La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb
36 36 36 36 36 36 36 36 36 36 36 36 36 36
Actinides Ac Th Pa U Np Pu Am Cm Bk Cf Es Fm Md No
68 68 68 68 68 68 68 68 68 68 68 68 68 68
%method FrozenCore FC_ELECTRONS # Freeze all core electrons

FC_EWIN # Freeze selected core electrons via
# %mp2 EWin EMin,EMax
FC_NONE # full MP2
NewNCore Bi 68 end # set the number of core electrons for
# Bi to 68
CheckFrozenCore true # check whether frozen core orbitals are
# ordered correctly
-n # freezes n electrons
end
NOTE
The FrozenCore options are applied to all post Hartree-Fock methods.
If including all electrons is desired, the !NoFrozenCore keyword can be simply inserted. For MP2:
Frozen virtual orbitals are not allowed in gradient runs or geometry optimization!
If ECPs are used, the number for NewNCore has to include the electrons represented by the ECPs as
well. E.g. if an element is supposed to have 60 electrons in the ECP and additional 8 electrons should
be frozen in the correlation calculation, NewNCore should be 68.
In ORCA we use rather conservative frozencore settings, i. e. a large number of electrons are included
in the correlation treatment. Therefore, we recommend to use properly optimized correlating functions
in all cases.
For DLPNO calculations the virtual space for core-core and core-valence correlation is adjusted by
default, which is described in detail in section 9.10.4.1.
In general, NewNCore only has an effect in calculations with FC ELECTRONS. In calculations using the
DLPNO approximation (except DLPNO-NEVPT2), NewNCore has also an effect in the other cases, as
is described in section 9.10.4.1.
9.9 The Second Order Many Body Pertubation Theory Module (MP2)
Throughout this section, indices i, j, k, . . . refer to occupied orbitals in the reference determinant, a, b, c, . . . to
virtual orbitals and p, q, r, . . . to general orbitals from either set while , , , , . . . refer to basis functions.
9.9.1 Standard MP2
The standard (or full accuracy) MP2 module has two different branches. One branch is used for energy
calculations, the other for gradient calculations.
For standard MP2 energies, the program performs two half-transformations and the half-transformed integrals
are stored on disk in compressed form. This appears to be the most efficient approach that can also be used
for medium sized molecules.The module should parallelize acceptably well as long as I/O is not limiting.
For standard MP2 gradients, the program performs four quarter transformations that are ordered by occupied
orbitals. Here, the program massively benefits from large core memory (%maxcore) since this minimizes the
number of batches that are to be done. I/O demands are minimal in this approach.
In memory mode (Q1Opt>0) basically the program treats batches of occupied orbitals at the same time.
Thus, there must be at least enough memory to treat a single occupied MO at each pass. Otherwise the MP2
module will fail. Thus, potentially, MP2 calculations on large molecules take significant memory and may be
most efficiently done through the RI approximation.
Alternatively, in the disk based mode (Q1Opt=-1) the program performs a half transformation of the
exchange integrals and stores the transformed integrals on disk. A bin-sort then leads to the AO operator
Kij (, ) = (i|j) in (11|22) integral notation. These integrals are then used to make the final Kij (a,b)
(a,b=virtual MOs) and the EMP2 pair energy contributions. In many cases, and in particular for larger
molecules, this algorithm is much more efficient than the memory based algorithm. It depends, however,
much more heavily on the I/O system of the computer that you use. It is important, that the program uses
the flags CFLOAT, UCFLOAT, CDOUBLE or UCDOUBLE in order to store the unsorted and sorted AO exchange
integrals. Which flag is used will influence the performance of the program and to some extent the accuracy
of the result (float based single precision results are usually very slightly less accurate; Eh-range deviations
from the double precision result14 ). Finally, gradients are presently only available for the memory based
algorithm since in this case a much larger set of integrals is required.
The ! MP2 command does the following: (a) it changes the Method to HFGTO and (b) it sets the flag DoMP2 to
true. The program will then first carry out a Hartree-Fock SCF calculation and then estimate the correlation
energy by MP2 theory. RHF, UHF and high-spin ROHF reference wavefunctions are permissible and the type
of MP2 calculation to be carried out (for high-spin ROHF the gradients are not available) is automatically
chosen based on the value of HFTyp. If the SCF is carried out conventionally, the MP2 calculation will also
14
However, sometimes, and in particular when transition metals and core orbitals are involved we have met unpleasantly
large errors. So be careful and double check when using floats!
9.9 The Second Order Many Body Pertubation Theory Module (MP2) 363
be done in a conventional scheme unless the user forces the calculation to be direct. For SCFMode=Direct or
SCFMode=SemiDirect the MP2 energy evaluation will be fully in the integral direct mode.
The following variables can be adjusted in the block for conventional MP2 calculations:
%mp2
EMin -1.5 # orbital energy cutoff that defines the
# frozen core in Eh
EMax 1.0e3 # orbital energy cutoff that defines the
# neglected virtual orbitals in Eh
EWin EMin,EMax # the same, but accessed as array
# (respects settings in %method block!)
MaxCore 256 # maximum amount of memory (in MB) to be
# used for integral buffering
ForceDirect false # Force the calculation to be integral
# direct
RI false # use the RI approximation
F12 false # apply F12 correction
Q1Opt # For non-RI calculations a flag how to perform
# the first quarter transformation
# 1 - use double precision buffers
# (default for gradient runs)
# 2 - use single precision buffers. This reduces
# the memory usage in the bottleneck step by
# a factor of two. If several passes are re-
# quired, the number of passes is reduced by
# a factor of two.
# -1 - Use a disk based algorithm. This respects
# the flags UCFLOAT,CFLOAT,UCDOUBLE and
# CDOUBLE. (but BE CAREFUL with FLOAT)
# (default for energy runs)
PrintLevel 2 # How much output to produce. PrintLevel 3 produces
# also pair correlation energies and other info.
DoSCS false # use spin-component scaling
Ps 1.2 # scaling factor for ab pairs
Pt 0.333 # scaling factor for aa and bb pairs
Densities none # no density construction
unrelaxed # only "unrelaxed densities"
relaxed # full relaxed densities
NatOrbs false # calculate natural orbitals
9.9.2 RI-MP2
The RI-MP2 module is of a straightforward nature. The program first transforms the three-index integrals
(ia|P ), where i is a occupied, a is a virtual MO and P is an auxiliary basis function that is orthogonalized
against the Coulomb metric. These integrals are stored on disk, which is not critical, even if the basis has
several thousand functions. The integral transformation is parallelized and has no specifically large core
memory requirements.
In the next step, the integrals are read ordered with respect to the occupied labels and the exchange operators
PNAux
K ij (a, b) = (ia|jb) = P (ia|P )(P |jb) are formed in the rate limiting O(N5 ) step. This step is done with
high efficiency by a large matrix multiplication and parallelizes well. From the exchange operators, the MP2
amplitudes and the MP2 energy is formed. The program mildly benefits from large core memory (%maxcore)
as this minimizes the number of batches and hence reads through the integral list.
The RI-MP2 gradient is also available. Here, all necessary intermediates are made on the fly.
In the RI approximation, one introduces an auxiliary fitting basis P (r) and then approximates the two-
electron integrals in the Coulomb metric as:
X
(pq|rs) (pq|P ) VP1
Q (Q|rs) (9.52)
PQ
where VP Q = (P |Q) is a two-index electron-electron repulsion integral. As first discussed by Weigend and
H
aser, the closed-shell case RI-MP2 gradient takes the form:
X X X
(x)
x
ERI-MP2 =2 (|P ) ci P
i +
x
VRS V1/2 V1/2 + hDFx i (9.53)
RS
P i RS
The F-matrix derivative terms are precisely handled as in the non-RI case and need not be discussed any
further. P
ia is a three-index two-particle density:
1/2
X
P
ia = (1 + ij ) tij
ab VP Q (Q|jb) (9.54)
jbQ
Which is partially transformed to the AO basis by:
X
P
i = ca P
ia (9.55)
a
The two-index analogue is given by:
1/2
X
P Q = Q
ia (ia|R) VRP (9.56)
iaR
The RI contribution to the Lagrangian is particularly convenient to calculate:

1/2
X X
LRI
ai = ca 2 P
i (|Q) VP Q (9.57)
P Q
In a similar way, the remaining contributions to the energy weighted density matrix can be obtained efficiently.
Note, however, that the response operator and solution of the CP-SCF equations still proceed via traditional
four-index integrals since the SCF operator was built in this way. Thus, while the derivatives of the three-index
integrals are readily and efficiently calculated, one still has the separable contribution to the gradient, which
requires the derivatives of the four-index integrals.
The RI-MP2 energy and gradient calculations can be drastically accelerated by employing the RIJCOSX or
the RIJDX approximation.
9.9.3 Double-Hybrid Density Functional Theory
A slightly more general form is met in the double-hybrid DFT gradient. The theory is briefly described
below.
The energy expression for perturbatively and gradient corrected hybrid functionals as proposed by Grimme
is:
Z Z
+ 1 (r1 ) (r2 ) 1 X
P P (| ) + cDF EXC [ , ] + cPT EPT

E = VN N + Ph + dr1 dr2 ax
2 |r1 r2 | 2
(9.58)
= ESCF + cPT EPT (9.59)
Here VN N is the nuclear repulsion energy and h is a matrix element of the usual one-electron operator which
contains the kinetic energy and electron-nuclear attraction terms (habi denotes the trace of the matrix product
ab). As usual, the molecular spin-orbitals are expanded in atom centered basis functions ( = , ):
X
p (r) = cp (r) (9.60)

with MO coefficients cp . The total density is given by (real orbitals are assumed throughout):
X 2
X
(r) = |i (r)| =
P (r) (r) = (r) + (r) (9.61)
i
Where P = P + P and P

ci ci .
P
= i
The second term of eq. 9.59 represents the Coulombic self-repulsion. The third term represents the contribution
of the Hartree-Fock exchange with the two-electron integrals being defined as:
Z Z
1
(| ) = (r1 ) (r1 ) r12 (r2 ) (r2 ) dr1 dr2 (9.62)
The mixing parameter ax controls the fraction of Hartree-Fock exchange and is of a semi-empirical nature.
The exchange correlation contribution may be written as:
GGA GGA
EXC [ , ] = (1 ax ) EX [ , ] + bEC [ , ] (9.63)
GGA GGA
Here, EX [ , ] is the exchange part of the XC-functional in question and EC [ , ] is the correlation
part. The parameter b controls the mixing of DFT correlation into the total energy and the parameter cDF
is a global scaling factor that allows one to proceed from Hartree-Fock theory (aX = 1, cDF = 0, cPT = 0)
to MP2 theory (aX = 1, cDF = 0, cPT = 1) to pure DFT (aX = 1, cDF = 0, cPT = 1) to hybrid DFT
(0 < aX < 1, cDF = 1, cPT = 0) and finally to the general perturbatively corrected methods discussed in this
work (0 < aX < 1, cDF = 1, 0 < cPT < 1). As discussed in detail by Grimme, the B2-PLYP functional uses
the Lee-Yang-Parr (LYP) functional as correlation part, the Becke 1988 (B88) functional as GGA exchange
part and the optimum choice of the semi-empirical parameters was determined to be aX = 0.53, cPT = 0.27,
cDF = 1, b = 1 cPT . For convenience, we will suppress the explicit reference to the parameters aX and b in
the XC part and rewrite the gradient corrected XC energy as:
Z
EXC , = f , , , , dr

(9.64)
0
~ 0 . The final term in eq (48) represents the scaled second order
~
with the gradient invariants =
perturbation energy:
1 X
i j i j + 1 X
i j i j + X
i j i j +
E PT2 = t K + t K + t K (9.65)
2 i <j 2 i <j i ,j

The PT2 amplitudes have been collected in matrices ti j0 with elements:
1
i j i j0 i + j 0 a b 0
ta b00 = K (9.66)
a b0
Where the orbitals were assumed to be canonical with orbital energies p . The exchange operator matrices are
i j i j0 = (i a |j0 b0 )
Ka b00 = (i a |j0 b0 ) and the anti-symmetrized exchange integrals are defined as K a b0
0 (i b | a ).
The orbitals satisfy the SCF equations with the matrix element of the SCF operator given by:
X

F = h + P (| ) aX P (| ) + cDF (|VXC |) (9.67)

The matrix elements of the XCpotential for a gradient corrected functional are: [59]
Z
f f ~ ~ f ~ ~
(|VXC |) = ( ) + 2 ( ) + ( ) dr (9.68)
(r)
The energy in equation 9.59 depends on the MO-coefficients, the PT2-amplitudes and through VN N , VeN
(in h) and the basis functions also explicitly on the molecular geometry. Unfortunately, the energy is only
stationary with respect to the PT2 amplitudes since they can be considered as having been optimized through
the minimization of the Hylleraas functional:

1 X
1 D E

X
X
i j + +
ti j K i j + +
ti j K i j + + D0 F+ + D0 F+
t i j K

EPT2 = min
t 2 2
i <j
i <j i j
(9.69)
The unrelaxed PT2 difference density is defined as:
0 1 X
i k k j X
i k k j
Dij = t t t t (9.70)
2
k k
0 X X
Dab = ti j ti j + + ti j + ti j (9.71)
i <j i j
With analogous expressions for the spin-down unrelaxed difference densities. Minimization of this functional
with respect to the amplitudes yields the second order perturbation energy. The derivative of the SCF part
of equation 9.59 with respect to a parameter x is straightforward and well known. It yields:
x x

P (x)
ESCF = VNx N + hPh
n i + WSCF S(x) + (| ) o
(x) ~ (x) (x)
X R
f ~ f ~
0~
f
+ (r) + 2 + 0 dr
(9.72)

|{z}
( 0 6=)
Superscript x refers to the derivative with respect to some perturbation x while a superscript in
parentheses indicates that only the derivative of the basis functions with respect to x is to be taken. For
example:
n o
(x)

+
P
= P x x

(9.73)
hx

= x |h| + | h |
+ |h| x x
In equation 9.72, S is the overlap matrix and WSCF the energy weighted density:
X
SCF ;SCF ;SCF
W = W + W = ci ci i (9.74)
i
At this point, the effective two-particle density matrix is fully separable and reads:
1 1 1
= P P ax P P ax P P (9.75)
2 2 2
The derivative of the PT2 part is considerably more complex, since EPT2 is not stationary with respect to
changes in the molecular orbitals. This necessitates the solution of the coupled-perturbed SCF (CP-SCF)
equations. We follow the standard practice and expand the perturbed orbitals in terms of the unperturbed
ones as:
X
p;x (r) = ;x
Uqp q (r) (9.76)
q
The occupied-occupied and virtual-virtual blocks of U are fixed, as usual, through the derivative of the
orthonormality constraints:
;x 1 (x)
Uij = Sij (9.77)
2
;x 1 (x)
Uab = Sab (9.78)
2
;x (x) ;x
Uia = Sia Uai (9.79)
(x) (x)
Where Spq = cp cq S . The remaining virtual-occupied block of Ux must be determined through
P
the solution of the CP-SCF equations. However, as shown by Handy and Schaefer, this step is unnecessary
and only a single set of CP-SCF equations (Z-vector equations) needs to be solved. To this end, one defines
the Lagrangian:
i j
L 0
(k i |j b ) tka jb
P P
ai = R (D )ai + 2 j b c (a c |j b ) tc b 2
j i P j k b j k (9.80)
+2 j b c (a c |j b ) tb c 2 j k b (k i |j b ) tb a
P
An analogous equation holds for Lai . The matrix elements of the response operator R (D0 ) are best evaluated
in the AO basis and then transformed into the MO basis. The AO basis matrix elements are given by:
0
R (D0 ) = h 2D0
P
(| ) D [(| ) + (k| )]
2 f 0 2
f ~ + 2 f ~ (D0 )
~ ( )
P R
+ (D ) ( ) + 2 P P
i
~ 0 + f
+ 2 f ~ 0 ~ ( ) dr
D D
where

(D0 ) = 0 0 0
D0 , D0 , (D ) , (D ) , (D ) (9.81)
The -gradient-parameters are evaluated as a mixture of PT2 difference densities and SCF densities. For
example:
(D0 ) = 2
~ 0
D
~ 0
P (9.82)
With
X 0

D0 (r) =

D (r) (r) (9.83)

X

P (r) =

P (r) (r) (9.84)

Having defined the Lagrangian, the following CP-SCF equations need to be solved for the elements of the
Z-vector:
(a i ) Zai

+ R (Z)ai = Lai (9.85)
The solution defines the occupied-virtual block of the relaxed difference density, which is given by:

D = D0 + Z (9.86)
For convenience, D is symmetrized since it will only be contracted with symmetric matrices afterwards.
After having solved the Z-vector equations, all parts of the energy weighted difference density matrix can be
readily calculated:
1 1 X X k j
Wij;PT2 = Dij i +
j R (D)ij (i a |k b ) tjakb (i a |k b ) tba (9.87)
2 2
k a b k a b
;PT2 1 X X j i
Wab = Dab (a +
b) (i a |j c ) tibjc (i a |j c ) tc b (9.88)
2 i j c i j c

j k
;PT2
X X
Wai = 2 (k i |j b ) tka jb 2 (k i |j b ) tb a (9.89)
j k b j k b
;PT2
Wia =
i Zai (9.90)
Once more, analogous equations hold for the spin-down case. With the relaxed difference density and energy
weighted density matrices in hand, one can finally proceed to evaluate the gradient of the PT2 part as
(WPT2 = W;PT2 + W;PT2 ):
(x)
= hDhxi + WPT2 S(x) + PT2
x

P
EPT2 (| )
Z r r r r
P f (x) f (x) f (x) (9.91)
+ (r) + 2 + dr
0

(6= 0 )
x x
The final derivative of eq. 9.59 is of course the sum ESCF + cP T EPT2 . Both derivatives should be evaluated
simultaneously in the interest of computational efficiency.
Note that the exchange-correlation contributions to the gradient take a somewhat more involved form than
might have been anticipated. In fact, from looking at the SCF XC-gradient (eq. 9.72) it could have been
(x) the
speculated that the PT2 part of the gradient is of the same form but with P being replaced by H,
relaxed PT2 difference density. This is, however, not the case. The underlying reason for the added complexity
apparent in equation 9.91 is that the XC contributions to the PT2 gradient arise from the contraction of
the relaxed PT2 difference density with the derivative of the SCF operator. Since the SCF operator already
contains the first derivative of the XC potential and the PT2 energy is not stationary with respect to changes
in the SCF density, a response type term arises which requires the evaluation of the second functional
derivative of the XC-functional. Finally, as is well known from MP2 gradient theory, the effective two-particle
density matrix contains a separable and a non-separable part:
PT2 NS
= D P D P D P + (9.92)
i j i j
X X X
i j
NS
= c
i ca cj c b ta b + ci ca cj cb ta b + 2 c
i ca cj c b ta b (9.93)
i j a b i j a b i j a b
Thus, the non-separable part is merely the back-transformation of the amplitudes from the MO to the AO
basis. It is, however, important to symmetrize the two-particle density matrix in order to be able to exploit
the full permutational symmetry of the AO derivative integrals.
9.9.4 Orbital Optimized MP2
The MP2 energy can be regarded as being stationary with respect to the MP2 amplitudes, since they can be
considered as having been optimized through the minimization of the Hylleraas functional:
n D E D Eo
0 + 1 | H
EMP2 = min 2 1 |H| 0 E0 |1 (9.94)
t
H is the 0th order Hamiltonian as proposed by Mller and Plesset, 0 is the reference determinant, 1
|HF i is the reference energy. The quantities t
is the first-order wave function and E0 = EHF = hHF | H
collectively denote the MP2 amplitudes.
The fundamental idea of the OO-MP2 method is to not only minimize the MP2 energy with respect to
the MP2 amplitudes, but to minimize the total energy additionally with respect to changes in the orbitals.
Since the MP2 energy is not variational with respect to the MO coefficients, no orbital relaxation due to
the correlation field is taken into account. If the reference determinant is poor, the low-order perturbative
correction then becomes unreliable. This may be alleviated to a large extent by choosing better orbitals in the
reference determinant. Numerical evidence for the correctness of this assumption will be presented below.
In order to allow for orbital relaxation, the Hylleraas functional can be regarded as a functional of the
wavefunction amplitudes t and the orbital rotation parameters R that will be defined below. Through
a suitable parameterization it becomes unnecessary to ensure orbital orthonormality through Lagrange
multipliers. The functional that we minimize reads:
|0 i + h1 | H
L {t, R} = E0 [R] + 2 h1 | H 0 E0 |1 i (9.95)
0 is the reference determinant. However, it does no longer correspond to the Hartree-Fock (HF) determinant.
Hence, the reference energy E0 [R] = h0 [R]| H |0 [R]i also changes during the variational process and is no
longer stationary with respect to the HF MO coefficients. Obviously, E0 [R] > EHF since the HF determinant
is, by construction, the single determinant with the lowest expectation value of the full Hamiltonian.
The reference energy is given as:
X 1X
E0 [R] = hi| h |ii + hij||iji (9.96)
i
2 ij
The first-order wave function excluding single excitations is:
1 X ij
tab | ab

|1 i = ij (9.97)
4
ijab
A conceptually important point is that Brillouins theorem [324] is no longer obeyed since the Fock matrix
will contain off-diagonal blocks. Under these circumstances the first-order wavefunction would contain
contributions from single excitations. Since the orbital optimization brings in all important effects of the
singles we prefer to leave them out of the treatment. Any attempt to the contrary will destroy the convergence
properties. We have nevertheless contemplated to include the single excitations perturbatively:
(2)
X |Fia |2
ESingles = (9.98)
ia
a i
The perturbative nature of this correction would destroy the stationary nature of the total energy and is hence
not desirable. Furthermore, results with inclusion of single excitation contributions represent no improvement
to the results reported below. They will therefore not be documented below and henceforth be omitted from
the OO-MP2 method by default.
The explicit form of the orbital-optimized MP2 Hylleraas functional employing the RI approximation
(OO-RI-MP2) becomes:
X
|ii + 1X X 0 X X
L [t, R] = hi|h hij||iji + (ia|P )iaP + Dij Fij + Dab Fab (9.99)
i
2 ij ij
iaP ab
with:
0 X X
iaP = VP1
Q (Q|jb)tij
ab (9.100)
Q jb
Z Z
1
(ia|P ) = i (r1 )a (r1 ) P (r2 )dr1 dr2 (9.101)
|r1 r2 |
Z Z
1
(P |Q) = p (r1 ) Q (r2 )dr1 dr2 (9.102)
|r1 r2 |
Here, {} is the set of orthonormal molecular orbitals and {} denotes the auxiliary basis set. Fpq denotes a
Fock matrix element:
X
|qi +
Fpq = hp| h hpk||qki (9.103)
k
and it is insisted that the orbitals diagonalize the occupied and virtual subspaces, respectively:
Fij = ij Fii = ij i
(9.104)
Fab = ab Faa = ab a
The MP2 like density blocks are,
jk
Dij = 21 kab tik
P
ab tab
ij (9.105)
Dab = 21 ijc tij
P
ac tbc
where the MP2 amplitudes in the case of a block diagonal Fock matrix are obtained through the condition L
tij
=
ab
0:
hij||abi
tij
ab = (9.106)
a + b i j
The orbital changes are parameterized by an anti-Hermitian matrix R and an exponential Ansatz,
cnew = cold exp(R)!

0 Ria (9.107)
R=
Ria 0
The orbitals changes to second order are,
exp(R) |ii = |ii + a Rai |ai 21 jb Rbi Rbj |ji + . . .

P P
(9.108)
exp(R) |ai = |ai i Rai |ii 12 jb Raj Rbj |bi + . . .
P P
Through this Ansatz it is ensured that the orbitals remain orthonormal and no Lagrangian multipliers need
to be introduced. The first-order expansion of the Fock operator due to the orbital rotations are:
X
(1)
Fpq [R] = Fpq [0] + Rpq + Rrp Frq [0] + Rrq Fpr [0] (9.109)
r
X
(1)
Rpq = Rck {hpc||qki + hpk||qci} (9.110)
kc

|qi , gpqrs hpq||rsi :
The first-order energy change becomes hpq hp| h
Rci (hci + hic ) + 12 ijc Rci (gcjij + gijcj ) + Rcj (gicij + gijic )
P P
L [t, R] = ic
0 0
)iaP 2 ikaP Rak (ik|P)iaP
P P
+2 iacP Rci (ac|P
(9.111)

P (1) P
ij Dij Rij + c (Rci Fcj + Rcj Fic )

P (1) P
+ ab Dab Rab k (Rak Fkb + Rbk Fak )

The condition for the energy functional to be stationary with respect to the orbital rotations LR
[t,R]
ai
= 0 ,
yields the expression for the orbital gradient and hence the expression for the OO-RI-MP2 Lagrangian.
L [t, R] X X
gai = 2Fai + 2 Dij Faj 2 Dab Fib + R(1) (D)ai (9.112)
Rai j
X X b
+2 (ac|P )0P
ia 2 (ik|P )0P
ia
cP kP
The goal of the orbital optimization process is to bring this gradient to zero. There are obviously many ways
to achieve this. In our experience, the following simple procedure is essentially satisfactory. We first build a
matrix B in the current MO basis with the following structure:
Bij = ij Fii
Bab = ab (Faa + ) (9.113)
Bai = Bia = gai
where is a level shift parameter. The occupied/occupied and virtual/virtual blocks of this matrix are
arbitrary but their definition has a bearing on the convergence properties of the method. The orbital energies
of the block diagonalized Fock matrix appear to be a logical choice. If the gradient is zero, the B-matrix is
diagonal. Hence one obtains an improved set of orbitals by diagonalizing B.
In order to accelerate convergence a standard DIIS scheme is used. [60, 178] However, in order to carry out
the DIIS extrapolation of the B-matrix it is essential that a common basis is used that does not change from
iteration to iteration. Since the B-matrix itself is defined in the molecular orbitals of the current iteration
we choose as a common set of orthonormal orbitals the MOs of the HF calculation. The extrapolation is
carried out in this basis and the extrapolated B-matrix is transformed back to the current set of MOs prior
to diagonalization. Obviously, the same strategy can be used for orbital optimization in any method for
which an orbital gradient is available.
For well behaved cases this simple scheme converges in 5-10 iterations. Transition metals and more complicated
molecules may require up to 20 iterations and level shifting in order to achieve convergence.
P
Upon convergence the sum of the matrix D and the density of the reference determinant P = i ci ci
form the true one-particle density matrix of the OO-MP2 approach that can be used for property or gradient
calculations.
9.9.5 RIJCOSX-RI-MP2 Gradients
Additional grids are introduced for the RIJCOSX-MP2 gradient. They have sensible default settings and
therefore do not usually require any intervention from the user. However, a number of expert options are
available, as described below.
The RIJCOSX-SCF procedure employs three grids: a small grid for the initial iterations, a medium grid
for the final iterations, and a large grid to evaluate the energy more accurately after the iterations have
converged. By default, the Z-vector equations are solved on a grid identical to the small grid in the SCF
procedure. However, any of the other SCF grids can be selected for the CPSCF procedure in the method
block as follows:
%method
Z_GridX 1 # small SCF grid (default)
2 # medium SCF grid
3 # large SCF grid
end
Yet another grid is used to evaluate basis functions derivatives. Appropriate parameters are chosen through
! GridXn (in addition to the three SCF grids), but there are several ways to override this setting. Firstly,
the grid can be chosen with ! MP2GridXn (n = 13), where the default is MP2GridX2. Secondly, the angular
(GridX) and radial (IntAccX) parameters can be specified explicitly through:
%mp2 GridX 2 # default 2: angular Lebedev grid 110

IntAccX 4.34 # radial grid
end
9.9.6 MP2 and RI-MP2 Second Derivatives
Analytical second-order properties with the MP2, RI-MP2 and double-hybrid DFT methods are available
in ORCA for calculations without frozen core orbitals. Hessians can be calculated with MP2 and RI-MP2.
The most expensive term in the second derivative calculations is the four-external contribution which can be
evaluated either via an AO direct (default) or a semi-numerical Chain-of-Spheres approach. In case that the
latter approach is chosen, appropriate grid parameters are defined through the ! GridXn settings. However,
a more fine-grained specification is available to expert users as follows:
%mp2 KCOpt _AOBLAS # (default) AO direct with BLAS routines

_COSX # semi-numerical evaluation using the COSX method
KC_GridX 2 # default 2: angular Lebedev grid 110
KC_IntAccX 3.34 # radial grid
end
Alternatively, all the grid settings can be defined in the %method block. The first three entries define the
three SCF grids, the fourth entry the MP2 grid for basis function derivatives (refer to section 9.9.5) and the
fifth entry the grid for the four-external contribution.
%method
IntAccX Acc1, Acc2, Acc3, Acc4, Acc5
GridX Ang1, Ang2, Ang3, Ang4, Ang5
UseFinalGridX true
end
9.9.7 Local MP2
In analogy to the domain-based local pair natural orbital coupled-cluster methods, there is also a local linear
scaling version of MP2 (DLPNO-MP2) implemented in ORCA. Its default thresholds are chosen to reproduce
about 99.9 % of the total RI-MP2 correlation energy, resulting in an accuracy of a fraction of 1 kcal/mol for
energy differences. The theory has been described in the literature. [75, 76]
Further information of local correlation methods in ORCA can be found in section 9.10.4. The local MP2
method becomes truly beneficial for very large molecules and can be used to compute energies of systems
containing several hundred atoms. Figure 9.2 shows the scaling behavior for linear alkane chains. Note that
this represents an idealized situation. For three-dimensional molecules the crossover with canonical RI-MP2
is going to occur at a later point.
Figure 9.2: Scaling of the DLPNO-MP2 method with default thresholds for linear alkane chains
in def2-TZVP basis. Shown are also the times for the corresponding Hartree-Fock
calculations with RIJCOSX and for RI-MP2.
In the following, the most important design principles of the RHF-DLPNO-MP2 are pointed out.
Unlike inqthe 2013 version of the DLPNO methodology, domains are selected by means of the differential
0 which are
R
overlap 02 (r)dr between localized MOs i and projected atomic orbitals (PAOs)
i2 (r)
normalized to unity. The default value for the corresponding cutoff is TCutDO = 102 .
MP2 amplitudes for each pair of localized orbitals ij are expressed in a basis of pair natural orbitals
(PNOs) aij . PNOs are obtained from diagonalization of an approximate, semicanonical MP2 pair
density Dij . Only PNOs with an occupation number > TCutPNO are retained, with a default value of
TCutPNO = 108 for DLPNO-MP2. The pair density is given by:
|j )
(i
ij + Tij T
ij Tij =
Dij = Tij T where + Fii Fjj (9.114)
ij = (1 + ij )1 4Tij 2Tij
T
Since the occupied block of the Fock matrix is not diagonal in the basis of localized orbitals, the
MP2 amplitudes Tij are obtained by solving the following set of residual equations iteratively (where
subscripts of PNOs have been dropped):
ij X X XX
Raij ij,kj kj kj,ij
Fkj Saij,ik ik ik,ij

b
= i
a j b + a
+ b Fii F jj Ta
b
Fik Sa c
T S
cd db c TcdSd
b
=0
k6=i cd k6=j cd
(9.115)
The largest part of the error relative to canonical RI-MP2 is controlled by these two thresholds,
which should be adequate for most applications. If increased accuracy is needed (e.g. for studying
weak interactions), tighter truncation criteria can be invoked by means of the ! TightPNO keyword.
Conversely, a less accurate but faster calculation can be performed with the ! LoosePNO keyword. For
more details refer to table 9.9.
Fitting domains are determined by means of orbital Mulliken populations with a threshold TCutMKN =
103 . This threshold results in an error contribution that is typically about an order of magnitude
smaller than the overall total energy deviation from RI-MP2.
Prior to performing the local MP2 calculation, pairs of localized molecular orbitals ij are prescreened
using an MP2 energy estimate with a dipole approximation, and the differential overlap integral
between orbitals i and j. This procedure has been chosen conservatively and leads to minimal errors.
Residual evaluation can be accelerated significantly by neglecting terms with associated Fock matrix
elements Fik and Fkj below FCut = 105 . Errors resulting from this approximation are typically below
1 Eh and thus negligible.
Sparsity of the MO and PAO coefficient matrices in atomic orbital basis is exploited to accelerate
integral transformations for large systems. Truncation of the coefficients is controlled by a parameter
TCutC . Neglect of these coefficients has to be performed very carefully in order to avoid uncontrollable
errors. The threshold has been chosen so as to make the errors essentially vanish.
By default, core orbitals are frozen in the MP2 module. However, if core orbitals are subject to an MP2
treatment, it is necessary to use a tighter PNO cutoff for pairs involving at least one core orbital. For
this purpose core orbitals and valence orbitals are localized separately. The cutoff for pairs involving
core orbitals is given by TCutPNO TScalePNO Core , where TScalePNO Core = 0.01 by default. For more
details refer to section 9.10.4.1.
The UHF-DLPNO-MP2 implementation differs somewhat from the RHF case, particularly regarding con-
struction of PNOs, as described below.
0 and
A separate set of PAOs 0 is obtained for each spin case.
For pairs, separate pair domains of PAOs need to be determined for each spin case. For example,
the pair domain [i j ] is the union of the domains [i ] and [j ] . The latter domain [j ] is
determined by evaluating the spatial differential overlap between j and -spin PAOs 0 .
One set of PNOs is needed for each same-spin pair. Opposite-spin pairs require a set of -PNOs and a
set of -PNOs. In total this results in four types of PNO sets.
Semicanonical amplitudes are obtained as follows, where i, j are spin orbitals and
are nonredundant
spin PAOs.
hij||
i
Tij = (9.116)
+ Fii Fjj
In the same-spin case hi j ||
i = hij|
i hij|
i, while in the opposite-spin case hi j ||
i =
hij|
i.
For opposite-spin pairs, -PNOs and -PNOs are obtained from diagonalisation of Tij Tij and Tij Tij ,
respectively. For same-spin pairs the pair density is symmetric and only one set of PNOs is needed.
PNOs are discarded whenever the absolute value of their natural occupation number if below the
threshold TCutPNO .
i j i j
The following residual equations need to be solved for the cases Rai jb , Ra b and Ra b :

D E
Rai jb = ij
ab + a + b Fii Fjj Taijb

X X X X (9.117)
Fik Saij,kj
c T kj kj,ij
S
cd d
b
Fkj Saij,ik ik ik,ij
c TcdSd
b
=0
k 6=i c d k 6=j c d
Most approximations are consistent between the RHF and UHF schemes, with the exception of the PNO
truncation. This means that results would match for closed-shell molecules with TCutPNO = 0 (provided
both Hartree-Fock solutions are identical), but this is not true whenever the PNO space is truncated.
Therefore, UHF-DLPNO-MP2 energies should only be compared to other UHF-DLPNO-MP2 energies,
even for closed-shell species.
We found it necessary to use tighter PNO thresholds for UHF-DLPNO-MP2. With NormalPNO settings
the default value is TCutPNO = 109 . For an overview of accuracy settings refer to table 9.9. As in the
RHF implementation, the PNO cutoff for pairs involving core orbitals is scaled with TScalePNO Core .
Table 9.9: Accuracy settings for DLPNO-MP2.
Setting TCutDO TCutPNO (RHF) TCutPNO (UHF)

LoosePNO 2 102 107 108
NormalPNO 1 102 108 109
TightPNO 5 103 109 1010
Options specific to DLPNO-MP2 are listed below.

%mp2 DLPNO false # Do DLPNO-MP2 (also requires RI true)

TolE 1e-7 # Energy convergence threshold
TolR 5e-7 # Residual convergence threshold
MaxPNOIter 25 # Maximum number of residual iterations
MaxLocIter 128 # Maximum number of iterations for orbital localization
LocMet FB # Localization method (options FB, PM, IAOIBO, IAOBOYS)
LocTol 1e-6 # Localization convergence tolerance
DIISStart_PNO 0 # First iteration to invoke DIIS extrapolation
MaxDIIS_PNO 7 # length of DIIS vector
Damp1_PNO 0.5 # Damping before DIIS is started
Damp2_PNO 0.0 # Damping with DIIS
MP2Shift_PNO 0.2 # level shift in amplitude update (Eh)
# Truncation parameters:
TCutPNO 1e-8 # PNO occupation number cutoff (RHF)
1e-9 # PNO occupation number cutoff (UHF)
TScalePNO_Core 1e-2 # Core PNO scaling factor
TCutDO 1e-2 # Differential overlap cutoff for domain selection
TCutMKN 1e-3 # Mulliken population cutoff for fitting domain selection
FCut 1e-5 # Occupied Fock matrix element cutoff
TCutPre 1e-6 # Energy threshold for dipole prescreening
TCutDOij 1e-5 # Maximum differential overlap between screened MOs
TCutDOPre 3e-2 # Cutoff to select PAOs for domains in prescreening
TCutC 1e-3 # Cutoff for PAO coefficient truncation
ScaleTCutC_MO 1.0 # Cutoff for MO truncation: TCutC * ScaleTCutC_MO
PAOOverlapThresh 1e-8 # Threshold for constructing non-redundant PAOs
end
9.10 The Single Reference Correlation Module
ORCA features a variety of single-reference correlation methods for single point energies (restricted to a
RHF or RKS determinant in the closed-shell case and a UHF or UKS determinant in the open-shell case;
quasi-restricted orbitals (QROs) [161] are also supported in the open-shell case). They are all fairly expensive
but maybe be used in order to obtain accurate results in the case that the reference determinant is a good
starting point for the expansion of the many-body wavefunction. The module is called orca mdci for matrix
driven configuration interaction. This is a rather technical term to emphasize that if one wants to implement
these methods (CCSD, QCISD etc.) efficiently, one needs to write them in terms of matrix operations which
pretty much every computer can drive at peak performance. Let us first briefly describe the theoretical
background of the methods that we have implemented in ORCA.
9.10.1 Theory
We start from the full CI hierarchy in which the wavefunction is expanded as:
9.10 The Single Reference Correlation Module 379
|i = |0i + |Si + |Di + |T i + |Qi + ... (9.118)
where |0i is a single-determinant reference and S, D, T, Q, . . . denote the single, double, triple quadruple
and higher excitations relative to this determinant at the spin-orbital level. As usual, labels i, j, k, l refer
to occupied orbitals in |0i, a, b, c, d to unoccupied MOs and p, q, r, s to general MOs. The action of the
second quantized excitation operators aai = aa ai on |0i lead to excited determinants |ai i that enter |i with
coefficients Cai . The variational equations are:
hai |H E0 | 0 + S + Di = EC Cai hai |H E0 | T i (9.119)
ij
ab

ab
ij |H E0 | 0 + S + D = EC Cab ij |H E0 | T + Q (9.120)
Further equations coupling triples with singles through pentuples etc.
The total energy is the sum of the reference energy E0 = h0 |H| 0i and the correlation energy
EC = h0 |H| S + Di (9.121)
which requires the exact singles- and doubles amplitudes to be known. In order to truncate the series to
singles- and doubles one may either neglect the terms containing the higher excitations on the right hand
side (leading to CISD) or approximate their effect thereby losing the variational character of the CI method
(CCSD, QCISD and CEPA methods). Defining the one- and two-body excitation operators as C1 = ia Cai aai ,
P
ij ab
C2 = 14 ijab Cab
P
aij one can proceed to approximate the triples and quadruples by the disconnected terms:
|T i = C1 C2 |0i (9.122)
1 2
|Qi = C |0i (9.123)
2 2
As is well known, the CCSD equations contain many more disconnected contributions arising from the various
powers of the C1 operator (if one would stick to CC logics one would usually label the cluster amplitudes with
tia , tij
ab ,. . . and the n-body cluster operators with Tn ; we take a CI point of view here). In order to obtain the
CEPA type equations from (9.119-9.123), it is most transparent to relabel the singles and doubles excitations
with a compound label P for the internal indices (i) or (ij ) and x for (a) or (ab). Then, the approximations
are as follows:
1 D x E 1 X D E
P (H E0 ) C22 0 = CyQ CzR xP |H E0 | yz (9.124)

QR
2 2
QRyz
X D E
CxP CyQ xP |H| xy
PQ (9.125)
Qy
X D E X D E
= CxP CyQ 0 |H| yQ CxP CyQ 0 |H| yQ (9.126)
Qy QyP x

X
CxP EC Q (9.127)
QP
Here the second line contains the approximation that only the terms in which either Qy or Rz are equal
to Px are kept (this destroys the unitary invariance) and the fourth line contains the approximation that
only exclusion principle violating (EPV) terms of internal labels are considered. The notation Qy P x
means Qy joint with Px (containing common orbital indices) and Q is the pair correlation energy. The
EPV terms must be subtracted from the correlation energy since they arise from double excitations that are
impossible due to the fact that an excitation out of an occupied or into an empty orbital of the reference
determinant has already been performed. Inserting eq. 9.127 into eq. 9.120 CxP EC cancels and effectively is
P
replaced by the partial correlation energy QP Q .
The resulting equations thus have the appearance of a diagonally shifted (dressed) CISD equation
hxP |H E0 + | 0 + S + Di = 0. If the second approximation mentioned above is avoided Malrieus
(SC)2 -CISD arises. [179, 280283] Otherwise, one obtains CEPA/3 with the shift:
X
ij
ab = (ik + jk ) ij (9.128)
k
CEPA/2 is obtained by ij ab = ij and CEPA/1 is the average of the CEPA/2 and CEPA/3. As mentioned
by Ahlrichs, [325] no consensus appears to exist in the literature for the appropriate shift on the single
excitations. If one proceeds straightforwardly in the same way as above, one obtains:
!
D E X
ai (H E0 ) C1 C2 0 Cai EC 2 ik (9.129)

k
as the appropriate effect of the disconnected triples on the singles. In has been assumed here that only
the singles |ai i in C1 contribute to the shift. If |0i is a HF determinant, the effect of the disconnected
triples in the doubles projection vanishes under the same CEPA approximations owing to Brillouins theorem.
Averaged CEPA models are derived by assuming that all pair correlation energies are equal (except ii = 0).
As previously discussed by Gdanitz [326], the averaging of CEPA/1 yields n2 EC and CEPA/3 EC n(n1) 4n6
where
n is the number of correlated electrons. These happen to be the shifts used for the averaged coupled-pair
functional (ACPF [327]) and averaged quadratic coupled-cluster (AQCC [328]) methods respectively. However,
averaging the singles shift of eq. 9.129 gives n4 EC . The latter is also the leading term in the expansion of the
AQCC shift for large n. In view of the instability of ACPF in certain situations, Gdanitz has proposed to
use the AQCC shift for the singles and the original ACPF shift for the doubles and called his new method
ACPF/2 [327]. Based on what has been argued above, we feel that it would be most consistent with the
ACPF approach to simply use n4 EC as the appropriate singles shift. We refer to this as NACPF.
It is readily demonstrated that the averaged models may be obtained by a variation of the modified correlation
energy functional:
h0 + S + D |H E0 | 0 + S + Di
EC = (9.130)
1 + gs hS|Si + gD hD|Di
with gS and gD being the statistical factors n4 , n2 , n(n1)

4n6
, as appropriate for the given method. Thus, unlike
the CEPA models, the averaged models fulfil a stationarity principle and are unitarily invariant. However, if
one thinks about localized internal MOs, it appears evident that the approximation of equal pair energies
must be one of rather limited validity and that a more detailed treatment of the electron pairs is warranted.
Maintaining a stationarity principle while providing a treatment of the pairs that closely resembles that of
the CEPA methods was achieved by Ahlrichs and co-workers in an ingenious way with the development of
the CPF method [80]. In this method, the correlation energy functional is written as:
D E
xp |H E0 | yQ

X xp |H| 0 X
EC = 2 + p (9.131)
NP NP NQ
Px P Qxy
with
X X
NP = 1 + TP Q (CyQ )2 (9.132)
Q y
The topological matrix for pairs P =(ij ) and Q = (kl) is chosen as: [226]
ik + il jk + jl
TP Q = + (9.133)
2ni 2nj
with ni being the number of electrons in orbital i in the reference determinant. The singles out of orbital
i are formally equated with P = (ii). At the spin-orbital level, ni = 1, for closed shells ni = 2. Using the
P
same topological matrix in P = Q TP Q Q one recovers the CEPA/1 shifts for the doubles in eq. 9.129. It
is straightforward to obtain the CPF equivalents of the other CEPA models by adjusting the TP Q matrix
appropriately. In our program, we have done so and we refer below to these methods as CPF/1, CPF/2 and
CPF/3 in analogy to the CEPA models (CPF/1 CPF). In fact, as discussed by Ahlrichs and co-workers,
variation of the CPF-functional leads to equations that very closely resemble the CEPA equation and can be
readily implemented along the same lines as a simple modification of a CISD program. Ahlrichs et al. argued
that the energies of CEPA/1 and CPF/1 should be very close. We have independently confirmed that in the
majority of cases, the total energies predicted by the two methods differ by less than 0.1 mEh.
An alternative to the CPF approach which is also based on variational optimization of an energy functional
is the VCEPA method [329]. The equations resulting from application of the variational principle to the
VCEPA functional are even closer to the CEPA equations than for CPF so that the resulting energies are
practically indistinguishable from the corresponding CEPA values. The VCEPA variants are referred to
as VCEPA/1, VCEPA/2, and VCEPA/3 in analogy to CEPA and CPF. A strictly size extensive energy
functional (SEOI) which is invariant with respect to unitary transformations within the occupied and virtual
orbital subspaces is also available [330] (an open-shell version is not implemented yet).
Again, a somewhat critical point concerns the single excitations. They do not account for a large fraction of
the correlation energy. However, large coefficients of the single excitations lead to instability and deterioration
of the results. Secondly, linear response properties are highly dependent on the effective energies of the singles
and their balanced treatment is therefore important. Since the CEPA and CPF methods amount to shifting
down the diagonal energies of the singles and doubles, instabilities are expected if the effective energy of an
excitation approaches the reference energy of even falls below it. In the CPF method this would show up
as denominators NP that are too small. The argument that the CPF denominators are too small has led
Chong and Langhoff to the proposal of the MCPF method which uses a slightly more elaborate averaging
than (NP NQ )1/2 [79].15 However, their modification was solely based on numerical arguments rather than
physical or mathematical reasoning. In the light of eq. 9.129 and the performance of the NACPF, it appears
to us that for the singles one should use twice the TP Q proposed by Ahlrichs and co-workers. The topological
matrix TP Q is modified in the following way for the (very slightly) modified method to which we refer to as
NCPF/1:
ik + il jk + jl
Tij,kl = + (9.134)
2ni 2nj
Tij,k = 0 (9.135)
ik + il
Ti,kl = 2 (9.136)
ni
Ti,k = 0 (9.137)
(note that TP Q =6 TQP for this choice). Thus, the effect of the singles on the doubles is set to zero based on
the analysis of the CEPA approximations and the effect of the singles on the singles is also set to zero. This
is a sensible choice since the product of two single excitations is a double excitation which is already included
in the SD space and thus none of them can belong to the outer space. It is straightforward to adapt this
reasoning about the single excitations to the CEPA versions as well as to NCPF/2 and NCPF/3.
The aforementioned ambiguities arising from the use of single excitations in coupled-pair methods can be
avoided by using correlation-adapted orbitals instead of Hartree-Fock orbitals thus eliminating the single
excitations. There are two alternatives: (a) Brueckner orbitals and (b) optimized orbitals obtained from the
variational optimization of the electronic energy with respect to the orbitals. Both approaches have already
been used for the coupled-cluster doubles (CCD) method [331, 332] and later been extended to coupled-pair
methods [333]. In the case of CCD, orbital optimization requires the solution of so-called (or Z vector)
equations [334]. There is, however, a cheaper alternative approximating the Z vector by a simple analytical
formula [335].
Furthermore, the parametrized coupled-cluster (pCCSD) method of Huntington and Nooijen [336], which
combines the accuracy of coupled-pair type methods for (usually superior to CCSD, at least for energies
and energy differences) with the higher stability of the coupled-cluster methods, is an attractive alternative.
Comprehensive numerical tests [337] indicate that particularly pCCSD(-1,1,1) (or pCCSD/1a) and pCCSD
(-1.5,1,1) (or pCCSD/2a) have great potential for accurate computational thermochemistry. These methods
can be employed by adding the simple keywords pCCSD/1a or pCCSD/2a to the first line of input. As
15
This method although it has been rather extensively used in the past is not implemented in ORCA. We
recommend to use our NCPF/1 instead.
mentioned in section 8.1.3.8, the LPNO variants of the pCCSD methods are also available for RHF and UHF
references via usage of the simple keywords LPNO-pCCSD/1a and LPNO-pCCSD/2a.
9.10.2 Closed-Shell Equations
Proceeding from spin-orbitals to the spatial orbitals of a closed-shell determinant leads to the actual working
equations of this work. Saebo, Meyer and Pulay have exploited the generator state formalism to arrive at
a set of highly efficient equations for the CISD problem [94]. A similar set of matrix formulated equations
for the CCSD and QCISD cases has been discussed by Werner and co-workers [338] and the MOLPRO
implementation is widely recognized to be particularly efficient. Equivalent explicit equations for the CISD
and CCSD methods were published by Scuseria et al. [339]16 The doubles equations for the residual vector
are (i 6 j, all a, b):
ij ij Pn kj
o P
ij kl
ab = Kab + K Cij ab + FV Cij + Cij FV ab ik
Fjk Cab + Fik Cab + Kkl Cab
k kl
2Cik Cik+ Kkj 12 Jkj + Kik 21 Jik 2Ckj Ckj+ ab
P
+
k
P 1 ik+ jk+ 1 ik kj+
2C J + 2J C + Jjk Cik + Ckj Jik+ ab
k (9.138)
P n ji k o
+Cai Fbj + Cbj Fai ij k
Kka Cb + Kkb Ca + Kia Cj + Kja Ci b
k
ij
ij Cab
The singles equations are:
P ik
kj
ai = Fai + FV Ci a Fij Caj ik
P
2Kjb Jjb Cba
j jkb
P

+ 2Kij Jij Cj + Fj 2Cij+ Cij + 2Kia Kia+ Cij+ a (9.139)
j
i Cai
The following definitions apply:
X ij
K Cij

ab
= (ac|bd) Ccd (9.140)
cd
pq
Krs = (pr|qs) (9.141)
pq
Jrs = (pq|rs) (9.142)
X
hABi = Apq Bqp (9.143)
pq
16
Our coupled-cluster implementation is largely based on this nice paper. The equations there have been extensively
verified to be correct.
The two-electron integrals are written in (11|22) notation and F is the closed-shell Fock operator with FV being
ij
its virtual sub-block. We do not assume the validity of Brillouins theorem. The amplitudes Cai , Cab have
been collected in vectors Ci and matrices Cij wherever appropriate. The shifts i and ij are dependent on
the method used and are defined in Table 9.10 for each method implemented in ORCA.
Table 9.10: Summary of the diagonal shifts used in various singles- and doubles methods discussed in this
chapter. The quantities i and ij are the correlation energy increments brought about by the
single- and the double excitations respectively. The partial denominators for the CPF type
methods Ni and Nij are specified in eq. 9.132.
Method Doubles Shift Singles Shift
CISD EC EC
CEPA/0 0 0
1 1 1 1
P P
CEPA/1 2 (i + j ) + 4 (ik + jk ) 2 ii + 2 ik
k k
CEPA/2 ij i + ij i + ii
(i + j ) ij i ij + 12
P P
CEPA/3 (ik + jk ) i + ik
k k
1
P P
NCEPA/1 4 (ik + jk ) ii + ik
k k
NCEPA/2 ij 2ii
1
P P
NCEPA/3 ij + 2 (ik + jk ) 2 ik
k k
1 i j 1
P ik jk 1 ii 1
P ik
CPF/1 Nij 2 ( Ni + + Nj ) +
4 (Nik Njk ) Ni 2 Nii + 2 Nik
n k o n k o
i i ii
CPF/2 Nij ij N i
+ Nijij Ni N + Nii
i
i P ik jk i ik
(1 ij ) + Njj Nijij + 12
P
CPF/3 Nij Ni ( Nik + Njk ) Ni N i + Nik
k k
1
P ik jk
4 Nij (N + ) ii P ik
NCPF/1 ik Njk Ni Nii + Nik
k k
ii
NCPF/2 Nij Nijij 2Ni Nii
P ik
P ik jk 2Ni
NCPF/3 Nij Nijij + 12 ( Nik + Njk ) Nik
k k
2 2
ACPF E
n C E
n C
h i
ACPF/2 2
n EC
1 (n3)(n2)
n(n1) EC
2 4
NACPF n EC n EC
h i h i
(n3)(n2)
AQCC 1 n(n1) EC 1 (n3)(n2)
n(n1) EC
The QCISD method requires some slight modifications. We found it most convenient to think about the
effect of the nonlinear terms as a dressing of the integrals occurring in equations 9.138 and 9.139. This
attitude is close to the recent arguments of Heully and Malrieu and may even open interesting new routes
towards the calculation of excited states and the incorporation of connected triple excitations. [340] The
dressed integrals are given by:
Fik = Fik + Cil 2Kkl Kkl+

(9.144)
l
X
Fab = Fab Ckl 2Kkl Kkl+ ab

(9.145)
kl
X
Fkc = Fkc + 2Kkl Kkl+ Cl

(9.146)
l
ij = K ij + Kkl Tkl+

K kl kl (9.147)
X
1

ij = K ij +
K Cik
Kkj
Kjk ki
+C K kj
(9.148)
ab ab
2 ab
k
ij ij
X
Jab Cki Kjk ab

= Jab + (9.149)
k
The CCSD method can be written in a similar way but requires 15 additional terms that we do not document
here. They may be taken conveniently from our paper about the LPNO-CCSD method [82].
A somewhat subtle point concerns the definition of the shifts in making the transition from spin-orbitals to
spatial orbitals. For example, the CEPA/2 shift becomes in the generator state formalism:

D
ab ij
E
ij 1 2 ij 1 1
ij | | = C + + C + (9.150)
ab
3 ij 3 ij ba
3 ij 3 ij
ab is a contravariant configuration state function, see Pulay et al. [334]. The parallel and antiparallel spin
( ij
pair energies are given by:
1 X h ij ij
i
ij ij

ij = Kab Kba Cab Cba (9.151)
2
ab
1 X ij ij

ij = Kab Cab (9.152)
2
ab
This formulation would maintain the exact equivalence of an orbital and a spin-orbital based code. Only
in the (unrealistic) case that the parallel and antiparallel pair correlation energies are equal the CEPA/2
shift of 9.10 arise. However, we have not found it possible to maintain the same equivalence for the CPF
method since the electron pairs defined by the generator state formalism are a combination of parallel and
antiparallel spin pairs. In order to maintain the maximum degree of internal consistency we have therefore
decided to follow the proposal of Ahlrichs and co-workers and use the topological matrix TP Q in equation
9.133 and the equivalents thereof in the CEPA and CPF methods that we have programmed.
9.10.3 Open-Shell Equations
We have used a non-redundant set of three spin cases (, , ) for which the doubles amplitudes are
optimized separately. The equations in the spin-unrestricted formalism are straightforwardly obtained from
the corresponding spin orbital equations by integrating out the spin. For implementing the unrestricted
QCISD and CCSD method, we applied the same strategy (dressed integrals) as in the spin-restricted case.
The resulting equations are quite cumbersome and will not be shown here explicitly [88].
Note that the definitions of the spin-unrestricted CEPA shifts differ from those of the spin-restricted formalism
described above (see Kollmar et al. [88]). Therefore, except for CEPA/1 and VCEPA/1 (and of course
CEPA/0), for which the spin-adaptation of the shift can be done in a consistent way, CEPA calculations of
closed-shell molecules yield slightly different energies for the spin-restricted and spin-unrestricted versions.
Since variant 1 is also the most accurate among the various CEPA variants [85], we recommend to use
variant 1 for coupled-pair type calculations. For the variants 2 and 3, reaction energies of reactions involving
closed-shell and open-shell molecules simultaneously should be calculated using the spin-unrestricted versions
only.
A subtle point for open-shell correlation methods is the choice of the reference determinant [341]. Single
reference correlation methods only yield reliable results if the reference determinant already provides a
good description of the systems electronic structure. However, an UHF reference wavefunction suffers from
spin-contamination which can spoil the results and lead to convergence problems. This can be avoided if
quasi-restricted orbitals (QROs) are used [161, 340] since the corresponding zeroth-order wavefunction is an
eigenfunction of the S2 operator and thus, no severe spin-contamination will appear. The coupled-pair and
coupled-cluster equations will be still solved in a spin-unrestricted formalism but the energy will be slightly
higher compared to the results obtained with a spin-polarized UHF reference determinant. Furthermore,
especially for more difficult systems like e.g. transition metal complexes, it is often advantageous to use
Kohn-Sham (KS) orbitals instead of HF orbitals.
9.10.4 Local correlation
As described in previous sections of the manual, ORCA features the extremely powerful LPNO and DLPNO
methods. LPNO stands for local pair natural orbital approximation and DLPNO for domain based
LPNO. These methods are designed to provide results as close as possible to the canonical coupled-cluster
results while gaining orders of magnitude of efficiency through a series of well-controlled approximations. In
fact, typically about 99.8% to 99.9% of the canonical correlation energy is recovered in such calculations.
Even higher accuracy is achievable but will ultimately come at much higher computational cost. The default
cut-offs are set such that the vast majority of chemically meaningful energy differences are reproduced to
better than 1 kcal/mol relative to the canonical results with the same basis set. Of the LPNO and DLPNO
methods, the LPNO is the older one and will eventually be discarded from ORCA. It has some higher order
scaling steps (up to N 5 ) while DLPNO is linear scaling and of similar accuracy. Amongst the DLPNO
methods, the first generation implementation of the DLPNO methods (DLPNO2013) is only near-linear
scaling, while the DLPNO implementation since ORCA 4.0 is linear scaling.
It is important to understand that the LPNO and DLPNO implementations are intimately tied to the RI
approximation. Hence, in these calculations one must specify a fitting basis set. The same rules as for
RI-MP2 apply: the auxiliary basis sets optimized for MP2 are just fine for PNO calculations. You can specify
several aux bases for the same job and the program will sort it out correctly.
The theory of the LPNO methods has been thoroughly described in the literature in a number of original
research papers. [82, 86, 89, 90]
Hence, it is sufficient to only point to a few significant design principles and features of these methods:
1. The correlation energy of any molecule can be written as a sum over the correlation energies of pairs of
electrons, labelled by the internal indices (ij) of pairs of orbitals that are occupied in the reference
determinant. If the orbitals (i) and (j) are localized, the pair correlation energy ij falls off very quickly
with distance, quite typically by about an order of magnitude per chemical bond that is separating
orbitals (i) and (j). Hence, by using a suitable cut-off for a reasonable pair correlation energy estimate
many electron pairs can be removed from the high-level treatment and only a linear scaling number of
electron pairs will make a significant contribution to the correlation energy.
2. The natural estimate for the pair correlation energy comes from second order many body perturbation
theory (MP2). However, canonical MP2 is scaling with the fifth power of the molecular size and is
hence not really attractive from a theoretical and computational point of view. Owing to the small
pre-factor RI-MP2 goes a long way to provide reasonably cheap estimates for pair correlation energies.
However, if one uses localized internal orbitals, then the MP2 energy expression must be cast in form
of linear equations. However, if one uses canonical virtual orbitals together with localized internal
orbitals and neglects the coupling terms coming from purely internal Fock matrix elements F(i,k) and
F(k,j) then one ends up with a fair approximation that is termed semi-canonical MP2 in ORCA. It
serves as a guess in the older LPNO method. For DLPNO this method is also not attractive.
3. In DLPNO, the guess is more sophisticated. Here the virtual spaces is spanned by projected atomic
orbitals (PAOs) that are assigned to domains of atoms that are associated with each local internal
orbital (i) and with the union of two such domains (i) and (j) for the electron pair (ij). One applies the
semi-local approximation, one obtains an excellent approximation to the semi-canonical MP2 energy.
This is called the semi-local approximation and it is linearly scaling with respect to computational
effort. If one iterates these equations to self-consistency to eliminate the coupling terms F(i,k) and
F(k,j) then one obtains the full local MP2 method (LMP2 or L-MP2). If the domains are made large
enough the results approach the canonical MP2 energy to arbitrary accuracy while still being linear
scaling with respect to computational resources. This method is the default for the DLPNO method.
4. Basically, the high-spin open-shell version of the DLPNO uses the same strategy as the closed-shell
variant to efficiently generate the open-shell PNOs in a consistent manner to the closed-shell formalism.
Through the development of the UHF-LPNO-CCSD method, we have realized that use of unrestricted
MP2 (UMP2) approach to define the open-shell PNOs introduces a few complexities; (1) the PNOs
for spin orbitals cannot be defined for - electron pairs and vice versa, (2) the diagonal PNOs
for singly occupied orbitals cannot be properly defined, and (3) the PNO space does not become
identical to that in the closed-shell LPNO framework in the closed-shell limit. However, to program
all the unrestricted CCSD terms in the LPNO basis, those PNOs are certainly necessary. Therefore,
in the UHF-LPNO-CCSD implementation, several terms, which, in many cases, give rather minor
contributions in the correlation energy are omitted. Due to these facts, the UHF-LPNO-CCSD does
not give identical results to that of RHF-LPNO-CCSD in the closed-shell limit. In addition, screening
of the weak pairs on the basis of the semi-canonical UMP2 pair-energy results in somewhat unbalanced
treatment of the closed- and open-shell states in some cases, leading to rather larger errors in the
reaction energies. To overcome those issues, in the high-spin open-shell DLPNO-CCSD method, the
PNOs are generated in the framework of semi-canonical NEVPT2 which smoothly converges to the
RHF-MP2 counterpart in the closed-shell limit. The open-shell DLPNO-CCSD, which is made available
from ORCA 4.0, includes a full set of the open-shell CCSD equations and is designed as a natural
extension to the RHF-DLPNO-CCSD.
5. Screening of the electron pairs according to a truncation parameter (in ORCA it is called TCutPairs ) is
not sufficient to obtain a highly performing local electronic structure method. The original work of
Pulay suggested to limit excitations out of the internal orbitals (i) and (j) to the domain associated
with the pair (ij). While this works well and has been implemented to perfection by Werner, Sch utz
and co-workers over the years, the pre-factor of such calculations is high, since the domains have to be
chosen large in order to recover 99.9% or more of the canonical correlation energy.
6. The ORCA developers have therefore turned to an approach that has been used with a high degree of
success in the early 1970s by Meyer, Kutzelnigg, Staemmler and their co-workers, namely the method
of pair natural orbitals (PNOs). As shown by Loewdin in his seminal paper from 1955, natural
orbitals (the eigenfunctions of the one-particle density matrix) provide the fastest possible convergence
of the correlated wavefunction with respect to the number of one-particle functions included in the
virtual space. It has been amply established that approximate natural orbitals are almost as succesful
as the true natural orbitals (which would require the knowledge of the exact wavefunction) in this
respect. While the success of approximate correlation treatments of many particle systems that use
approximate natural orbitals of the whole systems are somewhat limited, this is not the case for pair
natural orbitals. The latter have first been suggested as a basis for correlation calculations by England
and co-workers and, at the time, were given the name pseudonatural orbitals, a term that was used
by Meyer throughout his pioneering work.
7. The PNOs are approximate natural orbitals of a given electron pair. In order to generate them one
requires a one particle pair density matrix Dij the eigenfunctions of which are the PNOs themselves
while the corresponding eigenvalues are the PNO occupation numbers. While there are many creative
possibilities that can lead to slightly different PNO sets, a quite useful and natural approximation is
to generate such a density from the MP2 amplitudes as an expectation value (the unrelaxed MP2
density. One then uses a second threshold (in ORCA TCutPNO ) that controls the PNOs to be included
in the calculation. PNOs with an occupation number < TCutPNO are neglected.
8. PNOs of a given electron pair form an orthonormal set. However, PNOs belonging to different electron
pairs are not orthonormal and hence they overlap. This non-orthogonality leads to surprisingly few
complications because the PNOs stay orthogonal to all occupied orbitals. In practice, the equations for
PNO based correlation calculations are hardly more complex than the canonical equations.
9. The nice feature of these pair densities is that they become small when the pair interaction becomes
small. Hence weak pairs are correlated by very few PNOs. Therefore, the PNO expansion of the
wavefunction is extremely compact and there only is a linear scaling number of significant excitation
amplitudes that need to be considered.
10. A great feature of the PNOs is that they are self-adapting to the correlation situation that they are
supposed to describe. Hence, they are as delocalized as required by the physics and there is no ad-hoc
assumption about their location in space. However, it is clear that the PNOs are located in the same
region of space as the internal orbitals that they correlate because otherwise they would not contribute
to the correlation energy.
11. Capitalizing on this feature one can define generous domains and expand the PNOs in terms of the
PAOs and auxiliary fit functions (for the RI approximations) that are contained in these domains. In
ORCA this is controlled by the third significant truncation parameter TCutMKN . This is the basis of
the DLPNO method. In the older LPNO method, the PNOs are expanded in terms of the canonical
virtual orbitals and TCutMKN is only used for a local fit to the PNOs. In the linear-scaling DLPNO
implementation the domains expanding the PNOs in terms of the PAOs are controlled via TCutDO .
12. PNO expansion have the amazing advantage that the PNOs converge to a well-defined set as the basis
set is approaching completeness. Hence, the increase in the number of PNOs per electron pair is very
small upon enlargement of the orbital basis. In other words, correlation calculations with large basis
sets are not that much more expensive than correlation calculations with small basis sets. Thus, the
advantage of PNO over canonical calculations increases with the size of the basis set. This is a great
feature as large and flexible basis sets are a requirement for meaningful correlation calculations.
13. In summary, DLPNO and LPNO calculations are controlled by only three cut-off parameters with
well-defined meanings: a) TCutPairs , the cut-off for the electron pairs to be included in the coupled-pair
or coupled-cluster iterations, b) TCutPNO which controls how many PNOs are retained for a given
electron pair and c) TCutMKN that controls how large the domains are that the PNOs expand over. For
the linear-scaling DLPNO calculations the domain sizes are controlled via TCutDO .
14. It is clear that owing to the truncations a certain amount of error is introduced in the results. However,
having the LMP2 results available, one can compensate for the errors coming from TCutPairs and
TCutPNO . This is done in ORCA and the correction is added to the final correlation energy, thus
bringing it very close (to mEh accuracy or better) to the canonical result. TCutMKN is best dealt
with by making it conservative (at 1e3 to 1e4 the domains are about 2030 atoms large, which is
sufficient for an accurate treatment).
15. Note that the LPNO and DLPNO methods do not introduce any real space cut-offs. We refrain from
doing so and insist in our method development by making all truncations based on wavefunction or
energy parameters. We feel that this is the most unbiased approach and it involves no element of
chemical intuition or prejudice. Other researchers have decided differently and we do not criticize
their choices.
16. In the DLPNO method a highly efficient screening mechanism is operative. In this method one first
obtains a (quadratically scaling) multipole estimate for the pair correlation energy that is extremely
fast to compute (a few seconds, even for entire proteins). Only if this estimate is large enough, a given
electron pair is even considered for a LMP2 treatment. Quite typically pairs with energy contributions
of 1e6 Eh and smaller are very well described by the dipole-dipole estimate. Specifically, we drop pairs
with estimated energies < 0.01 TCutPairs and add their multipole energy sum to the final correlation
energy. These corrections tend to be extremely small, even for large molecules and are insignificant for
the energy. However, importantly, the multipole estimate ensures linear scaling in the MP2 treatment.
The pairs that then do not survive the pair-prescreening are called weak pairs in the ORCA or
DLPNO sense. They still play a role in the calculation of the triple excitation correction.
17. The calculation of triple excitation contributions is more involved and one does not have a perturbative
estimate available since the (T) contribution is perturbative itself. While the (T) contribution is
much smaller than the CCSD correlation energy, the errors introduced by the various local and PNO
approximations can be significant. We found that one has to include triples with at least one pair being
a weak LMP2 pair (with its LMP2 amplitudes) in order to arrive at sufficiently accurate results.
Given these explanations the various cut-off parameters that can be controlled in LPNO and DLPNO
calculations should be understandable and are listed below. We emphasize again that only the three
thresholds TCutPairs , TCutPNO and TCutMKN should be touched by the user, unless very specific questions
are addressed. The recommended way to control the accuracy of calculations is to specify TightPNO,
NormalPNO or LoosePNO keywords, rather than to change numeric values of cutoffs. Individual thresholds
should normally not be changed, as the defaults are sensible and lead to good cost/performance ratios.
%mdci TCutPairs 1e-4 # cut-off for the pair truncation

TCutPNO 3.33e-7 # cut-off for the PNO truncation
TCutDO 1e-2 # cut-off for the DLPNO domain construction
TCutMKN 1e-3 # cut-off for the local fit
# for DLPNO2013: also domain construction
# remaining options, tied to the three main cut-offs,
EXPERTS ONLY!
Localize 1 # flag for using localized orbitals
LocMet PM # Pipek Mezey (FB=Foster Boys is sensible too)
LocTol 1e-6 # Absolute threshold for the localization procedure
LocTolRel 1e-8 # Relative threshold for the localization procedure
LocMaxIter 128 # Maximum number of localization iterations
LocRandom 1 # default, take random seed for any localization
# For internal orbitals: choose best of 32 localizations
0 # take constant seed for any localization (for testing)
LocNAttempts np # number of localization attempts
# default: number of processes, minimum 8, if
# randomize true
# 1, if randomize false
# any number larger or equal np, if randomize true
PNONorm MP2Norm# default, old IEPANorm can also be used

(near identical results)
NrMP2Pairs_Trip 1 # number of MP2 pairs to be included in the triples
calculation
PAOOverlapThresh 1e-8 # generation of non-redundant PAOs from redundant ones
UseFullLMP2Guess true # Use iterative full LMP2 (for DLPNO)
SinglesFockUsePNOs true # compute the Singles Fock matrix (SFM) in PNOs.
# DLPNO2013: default for SinglesFockUsePNOs is false,
# by default RIJCOSX is used for the SFM, except when
# RIJK is given. In that case the RIJK-SFM is used.
LMP2MaxIter # max no of iterations in the MP2 equations
LMP2TolE 1e-7 # LMP2 energy convergence tolerance
LMP2TolR 1e-7 # LMP2 residual convergence tolerance
LMP2ScaleTCutPNO # PNO cutoff for LMP2 is: TCutPNO*LMP2ScaleTCutPNO

# Default: TCutPNO(DLPNO-MP2)/TCutPNO(DLPNO-CCSD) with
# respective TCutPNOs specific to Loose/Normal/TightPNO
LMP2FCut 1e-7 # LMP2 neglect cut-off for
off-diagonal Fock matrix elements
LMP2Damp 1.0 # Damping for LMP2 iterations
TCutTNO 0 # Cut-off for triples natural orbitals (0=automatic)
TCutDom -1 # -1= use TCutMKN to make domains
TCutPNOSingles -1 # -1= use 0.03*TCutPNO
TCutPreScr -1 # -1= use 0.01*TCutPairs for
multipole estimate based screening
TCutDeloc 0.1 # delocalization threshold for specification
of extended domains.
Necessary because PAOs are not strictly localized
TCutOSV 1e-6 # orbital specific virtuals used for pre-screening.
No critical
TScaleMP2Pairs 0.1
TScaleMKNStrong 10
TScaleMKNWeak 100
NOTE
Generally better DLPNO results are obtained when several runs of localization are undertaken using
different initial guesses. The different initial guesses are obtained using randomization (LocRandom).
However, randomization of the initial guess can lead to differently localized MOs in different calculations.
This can yield non-identical correlation energies, varying in the sub-kJ/mol range, for different runs on
the same machine.
In order to yield identical correlation energy results, randomization can be switched off (LocRandom
0). However, switching off randomization only leads to identical results on the same machine, but can
still lead to slightly different results (in the sub-kJ/mol range) on different machines.
Reproducibility of the correlation energy is expected to increase further if LocNAttempts is set to

higher values.
The input below shows how to perform a DLPNO calculation with settings that exactly reproduce the
canonical RI-MP2 result. They are not recommended for production use, but merely to show that if the local
approximations are pushed, then the result coincides with the canonical one. If one would set TCutPNO to
zero this would give canonical RI-CCSD. However, this is an absurdly inefficient calculation and hence not
done.
#
! def2-SV(P) def2/JK RI-JK DLPNO-CCSD VeryTightSCF RI-MP2
# obtain a result that only contains errors from the PNO approximation
# but no others
%mdci TCutPairs 0
TCutMKN 0
UseFullLMP2Guess true
LMP2FCut 1e-9
LMP2MaxIter 25
LMP2TolE 1e-10
LMP2TolR 1e-11
PAOOVerlapThresh 1e-9
end
! Bohrs
* xyz 0 1
C -1.505246952209632 1.048213673267046 -3.005665895986369
C 1.289678561934891 0.246429688933291 -3.259735682020124
C 2.834670835163566 1.157307360133605 -0.990383454919828
C 1.924119415395082 -0.128330938291771 1.465070676514038
C -0.931529472233802 -0.722841293992075 1.397639867298547
C -2.347670084056626 1.213332291655600 -0.217984867773136
H 2.084955694093313 0.973408301535989 -5.037750251258102
H 1.426532559234904 -1.831017720289521 -3.371063003813707
H -1.795307501459984 2.891278294563413 -3.927855043896308
H -2.709613973668925 -0.308515546176734 -4.026627646697411
H -4.404246093821399 0.941639912907262 -0.071175054238094
H -1.962867323232915 3.122079490952855 0.528101313545138
H -1.245096579039474 -2.621186110634707 0.594784162223769
H -1.699155144887690 -0.782162821007662 3.328959985756973
H 2.347109421287126 1.104305785540561 3.087624818244846
H 2.990679065503112 -1.888017241218143 1.775287572161196
H 4.862301668284708 0.796425411350593 -1.279131939569907
H 2.634027658640572 3.226752635113244 -0.827936424652650
*
9.10.4.1 Including core orbitals in the correlation treatment
In some chemical applications some or all of the chemical core electrons must be included in the correlation
treatment. In this case, it is necessary to tighten the TCutPNO thresholds for electron pairs in which chemical
core electrons are involved. This is now the default in DLPNO calculations.
For instance, one can decide to switch off the frozencore approximation and include all the electrons in the
correlation treatment. In this case, the program will use tighter thresholds by default for all electron pairs
and Singles that involve chemical core electrons. Note that, in this case, the use of properly optimized basis
functions for correlating the inner electrons is highly recommened.
! DLPNO-CCSD(T) def2-SVP def2-SVP/C NoFrozenCore

%mdci TSCALEPNO_CORE 0.01 # scaling factor for TCutPNO for electron pairs and
# Singles involving chemical core electrons
end
* xyz 0 1
Ti 0.0001595288 0.0000775041 0.0000000000
F 1.7595996122 0.0000634675 -0.0000000011
F -0.5865076471 1.6586935196 0.0000000018
F -0.5866248292 -0.8294172469 -1.4362516915
F -0.5866248311 -0.8294172443 1.4362516907
*
Another option is to choose the involved chemical core electrons by using an energy window. In this way all
electron pairs and Singles that involve chemical core electrons, which are in the defined energy window, are
affected by TScalePNO CORE.
! DLPNO-CCSD(T) def2-SVP def2-SVP/C

%method
FrozenCore FC_EWIN
end
%mdci
EWIN -40, 10000
end
* xyz 0 1
Ti 0.0001595288 0.0000775041 0.0000000000
F 1.7595996122 0.0000634675 -0.0000000011
F -0.5865076471 1.6586935196 0.0000000018
F -0.5866248292 -0.8294172469 -1.4362516915
F -0.5866248311 -0.8294172443 1.4362516907
*
A summary with the number of electrons affected by TScalePNO Core for the examples just discussed is
shown in Table 9.11.
Table 9.11: Number of chemical core electrons included in the DLPNO calculation and affected by
TScalePNO Core for the TiF4 examples
Chemical Core Electrons
Keyword Valence Electrons
Frozen Includeda
FrozenCore (default) 18 0 40
NoFrozenCore 0 18 40
EWIN -40, 10000 16 2 40
a using TScalePNO Core.

By default, ORCA provides a chemical meaningful definition for the number of electrons which belong to the
chemical core of each element. As already discussed, these default values define which pairs are affected by
TScalePNO Core. However, the user can modify the number of chemical core electrons for a specific element
via the NewNCore keyword.
! DLPNO-CCSD(T) def2-SVP def2-SVP/C NoFrozenCore

%method
NewNCore Ti 8 end
end
* xyz 0 1
Ti 0.0001595288 0.0000775041 0.0000000000
F 1.7595996122 0.0000634675 -0.0000000011
F -0.5865076471 1.6586935196 0.0000000018
F -0.5866248292 -0.8294172469 -1.4362516915
F -0.5866248311 -0.8294172443 1.4362516907
*
In the previous example, the number of chemical core electrons for Ti has been fixed to 8.
NOTE
Of course, if electrons are replaced by ECPs, they are not included in the correlation treatment.
If ECPs are used, the number for NewNCore has to include the electrons represented by the ECPs as
well. E.g. if an element is supposed to have 60 electrons in the ECP and additional 8 electrons should
be frozen in the correlation calculation, NewNCore should be 68.
The different sets of orbitals (chemical core electrons included in the correlation treatment and valence
electrons) are localized separately in order to avoid the mixing of core and valence orbitals.
9.10.5 Hilbert space multireference coupled-cluster approaches
MRCC branch of the MDCI module includes Hilbert space multireference coupled-cluster methods. These
are based on Jeziorski-Monkhorst ansatz of the wave operator [342], i.e. each of the reference configurations
is assigned its own cluster operator.
M

X
=
eT () | ih |, (9.153)
=1
where are reference configurations spanning a model space.
The cluster equations are then defined for each of the cluster operators. Their most expensive term is
basically the same as in single reference theory, except that it is calculated using cluster amplitudes of the
corresponding reference. The remaining coupling terms in MkCC or disconnected/linked terms in BWCC
have a n2v times lower scaling than the direct term. The energy is obtained by a diagonalization of effective
Hamiltonian
eff T ()
H = h()|H|()i + h()|HN ()e | iC . (9.154)
The methods are state-specific, so only one eigenvalue of effective Hamiltonian has a physical meaning.
At the moment, the following versions are implemented:
the MkCCSD [343]
BWCCSD [344] with a posteriori correction
LPNO-MkCCSD
LPNO-BWCCSD
MkCCSD(Tu ) [345] for closed-shell references
BWCCSD(T) [346] for closed-shell references
The number of reference configurations is not itself restricted. However, only up to mutually bi-excited
references can be included in the model space in the current implementation.
The truncation of cluster operator and/or the use of LPNO is specified like in the single reference case.
Furthermore, the user needs to specify the following keywords:
mrcc - flag to switch the HS MRCC

on = do a MRCC calculation
off (default) = single reference calculation
mrcctype - specifies the MRCC variant

mkcc for MkCC calculation
bwcc for BWCC
refs - sets the reference configurations

each reference is defined by a string of four characters: 2 (doubly occupied orbital), 0 (unoccupied
orbital), a (singly occupied orbital with spin ), b (singly occupied orbital with spin ). References
are separated by a comma. The size of the string of each reference can be reduced by the use of n docc
keyword (see below). We recommend that occupation of active orbitals is specified by refs keyword
and the number of internal (and core) orbitals is specified by n docc.
n docc - number of doubly occupied orbitals not specified in the refs keyword. We recommend that
occupation of active orbitals is specified by refs keyword and the number of internal (and core)
orbitals by n docc.
root - specifies the root (state) requested.

A non-negative integer smaller than the number of references specifies the state directly.
-1 stands for selection of the root based on the overlap of the eigenstate of effective Hamiltonian
with a given vector specified by root overlap keyword
default: 0 (ie. ground state)
root overlap - specifies the vector in the model space mentioned above.
Example: MkCCSD calculation of CH2 in cc-pVTZ basis set using a CAS(2,2) space
!ccsd cc-pvtz # selects the basis set and the truncation of cluster operator
%mdci
mrcc on # switches on the MRCC branch
mrcctype mkcc # selects MkCC
n docc 3 # there are three doubly occupied core/internal orbitals
refs "20,02" # specifies the two references in the CAS(2,2) space.
# Due to symmetry, only these two references are needed.
root -1 # select the root by overlap
root overlap "-0.9, 0.1" # -0.9 corresponds to "20" reference, 0.1 to "02"
end
9.10.6 The singles Fock term
In most MDCI calculations, there is an intermediate, which resembles closely to the SCF Fock matrix, and
similar methods are available to efficiently calculate it. In the followings, a short discussion will be given of
the so-called singles Fock term, which in the closed shell case has the form
X X
G(t1 )pq = tjb (2(pq|jb) (pj|qb)) = cp cq G(t1 ) , (9.155)
jb
The singles Fock matrix can be obtained via transformation from its counterpart (G(t1 ) ) in the atomic
orbital (AO) basis
X X
G(t1 ) = tjb (2(|jb) (j|b)) = P (t1 ) (2(| ) (| )), (9.156)
jb
where
X
P (t1 ) = tjb cj cb (9.157)
jb
is the analogue of the SCF density matrix for the singles Fock case. For the singles Coulomb (J(t1 ) ) case,
the density may be symmetrized (P (t1 ) = P (t1 ) + P (t1 ) ), and one may use the resolution of identity
approximation
X XX
1 1 1
J(t1 ) = P (t1 ) (| ) P (t1 ) (|r12 |A)VAB (B|r12 | ), (9.158)
AB
where A, B are elements of the RI/DF auxiliary fitting basis. Note that the factor of 2 in (9.156) is taken
care of by symmetrization. Since we are using a symmetric density, we may use the same efficient routine to
evaluate the singles Coulomb term as in the SCF case, see 9.2.2.3 and 9.2.2.4.
For the exchange case (K(t1 ) ), one possibility is to use the COSX approximation (see 9.2.2.6)
X X X X
K(t1 ) = P (t1 ) (| ) Qg A (rg ) P (t1 ) Xg , (9.159)
g
The COSX routine is able to deal with asymmetric densities as well, and thus, it can be used here similar to
the SCF case.
The other possibility is to use RI for exchange (RIK),

X X
1 1 1
K(t1 ) = cj C(t1 )j (| ) (j|r12 |A)VAB (B|r12 | j), (9.160)
j jAB
where
X
C(t1 )j = tjb cb , (9.161)
b
and the j is an orbital transformed using C(t1 ).
Using these approximations, there are two approximations for the total singles Fock term, RIJCOSX called
by the simple keyword RCSinglesFock and RIJK called by RIJKSinglesFock, see 8.1.3.1. For canonical
and LPNO methods, by default the program chooses the same approximation used in the SCF calculation.
DLPNO2013 uses RIJCOSX by default, while in DLPNO, the singles Fock term is also evaluated using PNOs
via SinglesFockUsePNOs, see 9.10.4. This behavior can also be changed by keywords in the method block.
%method RIJKSinglesFock 1 # 0 false, 1 true

RCSinglesFock 0 # 0 false, 1 true
end
9.10.7 Use of the MDCI Module
The MDCI module is fairly easy to use. The flags for the simple input lines have been described in section
6.2. The detailed listing of options is found below:
%mdci citype CISD # CI singles+doubles

QCISD # quadratic CI (singles+doubles)
CCSD # coupled-cluster singles+doubles
CEPA_1 # coupled-electron pair approximation 1
CEPA_2 #
CEPA_3 #
CPF_1 # Coupled-pair functional approximation 1
CPF_2 # (note that CPF/1 is identical with
CPF_3 # the original CPF of Ahlrichs et al.)
VCEPA_1 # Variational CEPA approximation 1
VCEPA_2 #
VCEPA_3 #
NCEPA_1 # our slightly modified versions of CEPA
NCEPA_2 # and CPF
NCEPA_3 #
NCPF_1 #
NCPF_2 #
NCPF_3 #
VNCEPA_1 #
VNCEPA_2 #
VNCEPA_3 #
ACPF # averaged coupled-pair functional
ACPF_2 # Gdanitz modification of it
NACPF # our modification of it
AQCC # Szalay + Bartlett
SEOI # a strictly size extensive energy functional
# maintaining unitary invariance (not yet
# available for UHF)
MP3 # MP3 calculation. With UseSCS=true it is
# SCS-MP3
ewin -3,1e3 # orbital energy window to determine which
# MOs are included in the treatment
# (respects settings in %method block)
Singles true # include single excitations in the
# treatment (default true)
Triples 0 # (T) correction in CCSD(T)/QCISD(T)
# default is no triples
1 # Algorithm 1 (lots of memory, fast)
2 # Algorithm 2 (less memory, about 2x slower)
Brueckner true # use Brueckner orbitals
# (default false)
Denmat none # no evaluation of density matrices
linearized # density matrix obtained by retaining
# only CEPA_0-like terms, i.e., those
# linear in the excitation amplitudes
unrelaxed # unrelaxed density matrices, i.e.,
# density matrices without orbital
# relaxation
orbopt # perform orbital optimization yielding
# fully relaxed density matrices (if
# citype chosen as CCSD or QCISD this option
# implies evaluation of the Z vector).
# (default: linearized)
ZSimple true # simplified evaluation of the Z vector
# in case of orbital optimized CCD
# (citype chosen as CCSD or QCISD and
# Denmat as orbopt) by using an
# analytical formula
false # explicit solution of Z vector
# equations
# in case of orbital optimized CCD
# (default: false)
UseQROs # use of quasi-restricted orbitals

# (default false)
Localize 0 # use localized MOs. Presently very little
# use is made of locality. It may help
# for interpretations. Localization is
# incompatible with the (T) correction
PM # Use Pipek-Mezey localized MOs
FB # use Foster-Boys localized MOs
NatOrbIters 0 # Perform natural orbital iterations.
# default is none. Not possible for CCSD
# and QCISD
pCCSDAB # the three parameters for parametrized
pCCSDCD # coupled-cluster (default is 1.0 which
pCCSDEF # corresponds to normal CCSD
# this defines how the rate limiting step is handled
# MO and AOX need lots of disk and I/O but if they
# can be done they are fast
KCOpt KC_MO # Perform full 4-index transformation
KC_AOBLAS# AO direct with BLAS (preferred)
KC_AO # AO direct handling of 3,4 externals
KC_RI # RI approximation of 3,4 externals
KC_RI2 # Alternative RI (not recommended)
KC_AOX # Do it from stored AO exchange integrals
PrintLevel 2 # Control the amount of output. For 3 and
# higher things like pair correlation
# energies are printed.
MaxIter 35 # Max. number of iterations
# How the integral transformation is done.
# Note that it is fine to do AOX or AO or AOBLAS
# together with trafo_ri
TrafoType trafo_jk # Partial trafo to J+K operators
trafo_ri # RI transformation of all
# integrals up to 2-externals
# (3-ext for (T))and rest on the
# fly
trafo_full # Full four index transformation.
# Automatically chosen for
# KCOpt=KC_MO
MaxCore 350 # Memory in MB - used for integral
# trafos and batching and for storage of
# integrals and amplitudes
# dont be too generous
STol 1e-5 # Max. element of the residual vector
# for convergence check
LShift 0.3 # Level shift to be used in update of
# coefficients
MaxDIIS 7 # Max number of DIIS vectors to be stored
# this lets you control how much and what is residing
# in central memory. May speed up things. Note that
# MaxCore is not respected here 9
InCore 0 # nothing in core
1 # + sigma-vector and amplitudes (default)
2 # + Jij(a,b) Kij(a,b) operators
3 # + DIIS vectors
4 # + 3-exernal integral Kia(b,c)
5 # + 4-external integrals Kab(c,d)
# this is identical to ALL
# the default is AUTO which means that incore
# is chosen based on MaxCore
end
9.11 The Complete Active Space Self-Consistent Field (CASSCF)

Module
9.11.1 General Description
The complete active space self-consistent field (CASSCF) method is a special form of a multiconfigurational
SCF method and can be thought of as an extension of the Hartree-Fock method. It is a very powerful method
to study static correlation effects and a solid basis for MR-CI and MR-PT treatments. It can be applied to
the ground state and excited states or averages thereof. The implementation in ORCA is fairly general and
reasonably efficient. However, CASSCF calculations are fairly complex and ultimately require a lot of insight
from the user in order to be successful. In addition to detailed description here, the manual explores some
typical examples in section 8.1.7.8. Furthermore, the manual is supplemented with a tutorial for CASSCF
that covers many practical tips on the calculation design and usage of the program.
The wavefunction. The wavefunction of a given CASSCF state is written as
S X
CkI Sk .

I = (9.162)
k

Here, SI is the CASSCF N -electron wavefunction for state I with total spin S. The set of Sk is a set
of configuration state functions (for example linear combination of Slater determinants) each adapted to a
total spin S. The expansion coefficients CkI represent the first set of variational parameters. Each CSF is
constructed from a common set of orthonormal molecular orbitals i (r) which are in turn expanded in basis
P
functions i (r) = ci (r). The MO coefficients ci form the second set of variational parameters.
The energy. The energy of the CASSCF wavefunction is given by the Rayleigh quotient
9.11 The Complete Active Space Self-Consistent Field (CASSCF) Module 401
D E
S
SI HBO I
E (c, C) =
S S , (9.163)
I I
and represents an upper bound to the true total energy. However, CASSCF calculations are not designed to
provide values for total energy which are close to the exact energy. The purpose of a CASSCF calculation is
to provide a qualitatively correct wavefunction, which forms a good starting point for a treatment of dynamic
electron correlation.
The CASSCF method is fully variational in the sense that the energy is made stationary with respect to
variations in both sets of MO and CI coefficients. At convergence, the gradient of the energy with respect to
the MO and CI coefficients vanishes
E (c, C)
= 0, (9.164)
ci
E (c, C)
= 0. (9.165)
CkI
Orbital spaces. In CASSCF calculations, the MO space is divided into three user defined subspaces:
The inactive orbitals are the orbitals which are doubly occupied in all configuration state functions
(labels i, j, k, l).
The active orbitals are the orbitals with variable occupation numbers in the various CSFs (labels
t, u, v, w).
The external orbitals (labels a, b, c, d)
Note that in older publications, the inactive and active orbitals are distinguished and referred to as internal
orbitals.
The wavefunction and energy is invariant with respect to unitary transformations within the three subspaces.
The special feature of a CASSCF wavefunction is that a fixed number of electrons is assigned to each subspace.
The internal subspace is of course completely filled but the CSFs in the active space constitute a full-CI of
n-electrons in m-orbitals. The CSF list is constructed such, however, that a wavefunction of well defined
total spin (and potential space) symmetry results. Such a wavefunction is referred to as a CASSCF(n,m)
wavefunction. The CSF list grows extremely quickly with the number of active orbitals and the number of
active electrons (basically factorially). Depending on the system, the limit of feasibility is roughly around
14 active orbitals or about one million CSFs in the active space. Larger active spaces are tractable with
approximate CI solver such as the Iterative-Configuration-Expansion CI (ICE-CI) described in 9.15 or the
Density Matrix Renormalization Group (DMRG) discussed in 9.13.17
Since the orbitals within the subspaces are only defined up to a unitary transformation, the program needs
to make some canonicalization choice.
17
For approximate full CI approaches, CASSCF is neither invariant to active-active rotations nor exactly size-
consistent.
In ORCA, the final orbitals by default are:
1. natural orbitals in the active space,
2. orbitals which diagonalize the CASSCF Fock matrix in the internal space and
3. orbitals which diagonalize the CASSCF Fock matrix in the external space.
State averaging. In many circumstances, it is desirable to optimize the orbitals not for a single state but
for the average of several states. In order to see what is done, the energy for state I is re-written as:
X X
EI (c, C) = p(I)
q hpq + pr(I)
qs (pq |rs ) (9.166)
pq pqrs
p(I) pr(I)
Here, q and qs are the one-and two-particle reduced electron density matrices for this state (labels
p, q, r, s span the internal and active subspaces):
p(I) = SI Eqp SI

q (9.167)
1
S p r
pr(I) I Eq Es qr Esp SI

qs = (9.168)
2
The average energy is simply obtained from averaging the density matrices using arbitrary weights wI that
are user defined but are constrained to sum to unity.
X
p(av)
q = wI qp(I) (9.169)
I
X
pr(av)
qs = wI pr(I)
qs (9.170)
I
X
wI = 1 (9.171)
I
Optimization of CASSCF wavefunctions. In general, except for trivial cases, CASSCF wavefunctions
are considerably more difficult to optimize than RHF (or UHF) wavefunctions. The underlying reason is that
variations in c and C maybe strongly coupled and the energy functional may have many local minima in (c,C)
space. Consequently, the choice of starting orbitals is of really high importance and the choice which orbitals
and electrons are included in the active space has decisive influence on the success of a CASSCF study. In
general, after transformation to natural orbitals, one can classify the active space orbitals by their occupation
numbers which vary between 0.0 and 2.0. In general, convergence problems are almost guaranteed if orbitals
with occupation numbers close to zero or close to 2.0 are included in the active space. Occupation numbers
between 0.02 and 1.98 are typically very reasonable and should not lead to large convergence problems.
The reason for the occurrence of convergence problems is that the energy is only very weakly dependent on
rotations between internal and active orbitals if the active orbital is almost doubly occupied and similarly for
the rotations between external and weakly occupied active orbitals. However, in some cases (for example in
the study of potential energy surfaces) it may not be avoidable to include weakly or almost inactive orbitals
in the active space and in these cases the use of the most powerful convergence aids is necessary (vide infra).
As in the case of single-determinant wavefunctions (RHF, UHF, RKS, UKS) there are first and second order
converging methods available. The first order CASSCF methods require the transformed integrals (tu|vx)
with x belonging to any subspace. This is a very small subspace of the total transformed integral list and
is readily held in central storage even for larger calculations. On the other hand, second order CASSCF
methods require the integrals (pq|xy) and (px|qy) (p, q = internal, active; x, y = any orbital). This is a fairly
large set of integrals and their generation is laborious in terms of CPU time and disk storage. Second order
CASSCF calculations are therefore more limited in the size of the molecules which can be well treated. It
would be possible to basically avoid the integral transformation also in the case of second-order CASSCF
calculations and proceed to fully direct calculations. Such calculations may become quite time consuming
since there may be a large number of Fock matrix builds necessary.
The augmented Hessian method (Newton-Raphson) solves the eigenvalue problem:

! ! !
0 g 1 1
= (9.172)
g H t t
Here, g is the orbital gradient (derivative of the total energy with respect to a non-redundant rotation between
two orbitals) and H is the orbital Hessian (second derivative of the energy with respect to two non-redundant
orbital rotations). The vector t (in intermediate normalization obtained from the CI like vector) summarizes
the rotation angles. The angles are used to define the antisymmetric matrix (Xpq = Xqp is thus the rotation
angle between orbitals p and q):
!
0 t
X= , (9.173)
t 0
which is used to parametrize the unitary matrix U = exp (X) which is used to update the orbitals according
to:
cnew = cold U (9.174)
(where c is an MO coefficient matrix).
Starting orbitals. You cannot be careful enough with your starting orbitals. What type of initial guess
works best, depends on the system. Quit often it is not the magnitude of the initial gradient, but the
similarity between initial and final orbitals.The CASSCF tutorial discusses a number of guess options in more
detail. Generally speaking, canonical orbitals or HF orbitals from a RHF calculation are not good choice!
Usually DFT orbitals (quasi-restricted or RKS) often do better. Alternatively, if CASSCF orbitals from a
previous run or a close-by geometry are available this is a good choice. In many instances, e.g. transition
metal complexes, the PATOM guess produces more reliable start orbitals than the PMODEL guess. However,
natural orbitals from a simple correlation calculation like MP2 or a calculation with the MRCI module are
usually the best choice. For example:
#
# First job provides reasonable natural orbitals
#
! RHF RI-MP2 SVP def2-SVP/C SmallPrint
%mp2 natorbs true

density relaxed
end
* int 0 1
C 0 0 0 0.00 0.0 0.00
O 1 0 0 1.20 0.0 0.00
H 1 2 0 1.10 120.0 0.00
H 1 2 3 1.10 120.0 180.00
*
Now examine the occupation numbers of the natural orbitals (you will find that in the output of the MP2
part of the calculation):
Natural Orbital Occupation Numbers:

N[ 0] = 2.00000622
N[ 1] = 2.00000233
N[ 2] = 1.98685690
N[ 3] = 1.97743095
N[ 4] = 1.97533829
N[ 5] = 1.96915811
N[ 6] = 1.96443726
N[ 7] = 1.93992919
N[ 8] = 0.05371120
N[ 9] = 0.02543285
N[ 10] = 0.02444375
N[ 11] = 0.01355852
N[ 12] = 0.01093080
N[ 13] = 0.00984154
N[ 14] = 0.00648238
N[ 15] = 0.00626053
A rule of thumb is that orbitals with occupation numbers between 1.98 and 0.02 should be in the active space.
Thus, in the present case we speculate that a 10 electrons in 8 orbitals active space would be appropriate for
the CASSCF of the ground state. Lets try:
#
# Run a CASSCF calculation for the ground state of H2CO
#
! RHF SVP def2-SVP/C SmallPrint
! moread
%moinp "Test-CASSCF-MP2-H2CO.mp2nat"
%casscf nel 10
norb 8
mult 1
switchstep nr
end
* int 0 1
C 0 0 0 0.00 0.0 0.00
O 1 0 0 1.20 0.0 0.00
H 1 2 0 1.10 120.0 0.00
H 1 2 3 1.10 120.0 180.00
*
If we run that calculation it converges (somewhat a little slow in 8 iterations) and produces the following:

E(CAS)= -113.889220348 Eh DE= -0.000018747
N(occ)= 1.99752 1.99621 1.98487 1.97919 1.94279 0.05943 0.02156 0.01843
||g|| = 0.000764063 Max(G)= 0.000415430 Rot=9,1
From which we see that we had two orbitals too many in the active space with occupation numbers very close
to two. The presence of barely correlated orbitals (occupation close to 0.0 or 2.0) can cause convergence
problems. Their inclusion in the active space does not significantly change the energy and it might better to
omit these orbitals from the start.
In the present case, we re-run the CASSCF with 6 active electrons in six orbitals. The result is:
MACRO-ITERATION 2:
E(CAS)= -113.881917477 Eh DE= -0.010367388
N(occ)= 1.98207 1.97921 1.94013 0.06065 0.02088 0.01705
||g|| = 0.041684361 Max(G)= -0.017003745 Rot=5,4
--- SX_PT (Skipped TA=0 IT=0): ||X|| = 0.186646455 Max(X)(5,4) = 0.167228161

MACRO-ITERATION 3:
E(CAS)= -113.884702316 Eh DE= -0.002784839
N(occ)= 1.98170 1.97906 1.94109 0.05960 0.02110 0.01745
||g|| = 0.016234253 Max(G)= 0.008518352 Rot=10,4
MACRO-ITERATION 4:
E(CAS)= -113.885056475 Eh DE= -0.000354159
N(occ)= 1.98171 1.97917 1.94173 0.05891 0.02102 0.01746
||g|| = 0.010311921 Max(G)= -0.005579778 Rot=9,2
MACRO-ITERATION 5:
E(CAS)= -113.885235194 Eh DE= -0.000178719
N(occ)= 1.98175 1.97918 1.94210 0.05850 0.02105 0.01742
||g|| = 0.006833530 Max(G)= -0.003968813 Rot=9,2
MACRO-ITERATION 6:
E(CAS)= -113.885360841 Eh DE= -0.000125646
N(occ)= 1.98176 1.97925 1.94222 0.05834 0.02103 0.01741
||g|| = 0.004409607 Max(G)= -0.003128377 Rot=9,2
MACRO-ITERATION 7:
E(CAS)= -113.885406351 Eh DE= -0.000045510

N(occ)= 1.98169 1.97931 1.94207 0.05845 0.02101 0.01746
||g|| = 0.002225744 Max(G)= -0.000966988 Rot=10,4
MACRO-ITERATION 8:
E(CAS)= -113.885407435 Eh DE= -0.000001084
N(occ)= 1.98171 1.97932 1.94199 0.05854 0.02100 0.01745
||g|| = 0.000779783 Max(G)= -0.000296797 Rot=9,2
MACRO-ITERATION 9:
E(CAS)= -113.885407435 Eh DE= 0.000000000
N(occ)= 1.98173 1.97932 1.94206 0.05846 0.02100 0.01742
||g|| = 0.000779783 Max(G)= 0.000296787 Rot=9,2
--------------
CASSCF RESULTS
--------------
Final CASSCF energy : -113.885407435 Eh -3098.9795 eV
The calculation converges very quickly and the occupation numbers show you that all of these orbitals are
actually needed in the active space. The omission of the two orbitals from the active space came at an
increase of the energy by 4 mEh which seems to be tolerable. Lets look what we have in the active space
in figure 9.3.
Thus, we can see that we got a fairly nice result: our calculation has correlated the in-plane oxygen lone
pair, the C-O and the C-O bond. For each strongly occupied bonding orbital, there is an accompanying
weakly occupied antibonding orbital in the active space that is characterized by one more node. In particular,
the correlating lone pair and the C-O orbital would have been hard to find with any other procedure than
the one chosen based on natural orbitals. We have now done it blindly and looked at the orbitals only after
the CASSCF a better approach is normally to look at the starting orbitals before you enter a potentially
expensive CASSCF calculation. If you have bonding/antibonding pairs in the active space plus perhaps the
singly-occupied MOs of the system you probably have chosen a reasonable active space.
We can play the game now somewhat more seriously and optimize the geometry of the molecule using a
reasonable basis set:
(a) MO5 (b) MO6 (c) MO7
(d) MO10 (e) MO9 (f) MO8
Figure 9.3: Orbitals of the active space for the CASSCF(6,6) calculation of H2 CO.
! def2-TZVP def2-TZVP/C SmallPrint Opt

! moread
%moinp "Test-CASSCF-MP2-H2CO.mp2nat"
%casscf nel 6
norb 6
switchstep nr
end
* int 0 1
C 0 0 0 0.00 0.0 0.00
O 1 0 0 1.20 0.0 0.00
H 1 2 0 1.10 120.0 0.00
H 1 2 3 1.10 120.0 180.00
*
and get:
---------------------------------------------------------------------------


----------------------------------------------------------------------------
1. B(O 1,C 0) 1.2101 0.000259 -0.0002 1.2100
2. B(H 2,C 0) 1.0942 -0.000029 0.0001 1.0943
3. B(H 3,C 0) 1.0942 -0.000029 0.0001 1.0943
4. A(O 1,C 0,H 3) 122.07 0.000023 -0.00 122.07
5. A(H 2,C 0,H 3) 115.85 -0.000046 0.01 115.86
6. A(O 1,C 0,H 2) 122.07 0.000023 -0.00 122.07
7. I(O 1,H 3,H 2,C 0) -0.00 -0.000000 0.00 -0.00
----------------------------------------------------------------------------
Let us compare to MP2 geometries (this job was actually run first):
! RHF RI-MP2 def2-TZVP def2-TZVP/C SmallPrint TightSCF Opt
%mp2 natorbs true

end
* int 0 1
C 0 0 0 0.00 0.0 0.00
O 1 0 0 1.20 0.0 0.00
H 1 2 0 1.10 120.0 0.00
H 1 2 3 1.10 120.0 180.00
*
-------------------------------------------------------------------------


----------------------------------------------------------------------------
1. B(O 1,C 0) 1.2127 0.000374 -0.0002 1.2125
2. B(H 2,C 0) 1.0991 -0.000031 0.0001 1.0992
3. B(H 3,C 0) 1.0991 -0.000031 0.0001 1.0992
4. A(O 1,C 0,H 3) 121.77 0.000023 -0.00 121.77
5. A(H 2,C 0,H 3) 116.45 -0.000046 0.01 116.46
6. A(O 1,C 0,H 2) 121.77 0.000023 -0.00 121.77
7. I(O 1,H 3,H 2,C 0) -0.00 -0.000000 0.00 -0.00
----------------------------------------------------------------------------
The results are actually extremely similar (better than 1 pm agreement). Compare to RHF:
---------------------------------------------------------------------------


----------------------------------------------------------------------------
1. B(O 1,C 0) 1.1784 -0.000164 0.0001 1.1785
2. B(H 2,C 0) 1.0921 0.000010 -0.0000 1.0921
3. B(H 3,C 0) 1.0921 0.000010 -0.0000 1.0921
4. A(O 1,C 0,H 3) 121.93 -0.000003 -0.00 121.93
5. A(H 2,C 0,H 3) 116.13 0.000005 0.00 116.13
6. A(O 1,C 0,H 2) 121.93 -0.000003 -0.00 121.93
7. I(O 1,H 3,H 2,C 0) 0.00 0.000000 -0.00 -0.00
----------------------------------------------------------------------------
Thus, one can observe that the correlation brought in by CASSCF or MP2 has an important effect on the
C=O distance (4 pm), while the rest of the geometry is not much affected.
More on the technical use of the CASSCF program.
The most elementary input information which is always required for CASSCF calculations is the specification
of the number of active electrons and orbitals.
%casscf nel 4 # number of active space electrons

norb 6 # number of active orbitals
end
The CASSCF program in ORCA can average states of several multiplicities. The multiplicities are given as a
list. For each multiplicity the number of roots should be specified:
%casscf mult 1,3 # here: multiplicities singlet and triplet
nroots 4,2 # four singlets, two triplets

end
If the symmetry handling in ORCA is enabled (! UseSym) each multiplicity block must have an irreducible
representation assigned. Numbers corresponding to the irrep within a given symmetry are printed in the
output of ORCA.

irrep 0,1 # here: irrep for each mult. block (mandatory!)

end
Several roots and multiplicities usually imply a state average CASSCF (SA-CASSCF) calculation. The
program by default chooses equal weights for the multiplicity blocks. Roots within a given block have equal
weight. Users can define a custom weighting scheme for the multiplicity blocks and roots:

bweight 2,1 # singlets and triplets weighted 2:1
weights[0] = 0.5,0.2,0.2,0.2 # singlet weights
weights[1] = 0.7,0.3 # triplet weights
end
The program automatically normalizes these weights such that the sum over all weights is unity. If convergence
on an excited state is desired then the weights[0] array may look like 0.0,0.0,1.0 (this would optimize the
orbitals for the third excited state. If several states cross during the orbital optimization this will ultimately
cause convergence problems.
Orbital optimization methods. In the following we discuss the available options for orbital optimization.
These keywords are optional and should only be used when facing convergence difficulties. We
have implemented several orbital optimization methods (list below).
# Keywords to be used as Orbstep/Switchstep

SuperCI_PT # perturbative SuperCI (first order)
SuperCI # SuperCI (first order)
DIIS # DIIS (first order)
KDIIS # KDIIS (first order)
SOSCF # approx. Newton-Raphson (first order)
NR # augmented Hessian Newton-Raphson
# unfolded two-step procedure
# - still not true second order
The different convergers have different strengths. First order method are cheap but typically require more
iterations compared to second order methods. When the gradient is far off from convergence the program
uses the converger defined as orbstep while close to convergence the switchstep is used. The actual criteria
for switchstep are defined with the keywords SwitchConv and SwitchIter.
%casscf
OrbStep SuperCI # or any other from the list above
SwitchStep DIIS # or any other from the list above
SwitchConv 0.03 # gradient at which to switch

SwitchIter 15 # iteration at which the switch takes place
# irrespective of the gradient
MaxIter 75 # Maximum number of macro-iterations

end
Picking a convergence strategy, the program has to balance speed and robustness. The default strategy uses
the SuperCI PT as converger for orbstep and switchstep. This approach determines the elements Xpq of
the anti-Hermitean matrix used in the orbital update according to
Cnew = Cold eX (9.175)
from first order perturbation theory using the the Dyall-Hamiltonian [347] in zeroth order and a first-order
perturbed wave function given as (1) = pq qp Xqp where the qp represent singly excited functions obtained
P
from the CASSCF wave function by excitation from orbital p to orbital q . The SuperCI PT is robust with
respect to orbitals that are exactly doubly occupied or empty. Rotations with orbital close to this critical
occupations can further be eliminated with the keyword DThresh (default=1e-6). However, the method
is quiet aggressive in the orbital optimization. In some cases, such as basis set projection or PATOM guess
(intrinsic basis set projection), the program might pick a step-size that is too big. Then restricting the
step-size via the keyword MaxRot (default=0.2) might be useful. The keywords DThresh and MaxRot are
specific to SuperCI PT. In contrast to other convergers, level shift and other means of damping
do not apply to the SuperCI PT.
MaxRot 0.05 # cap stepsize for SuperCI_PT

DThresh 1e-6 # thresh for critical occupation
Previous default settings used the combination of orbstep SuperCI and switchstep DIIS, which is more
robust than the present default (SuperCI PT), but typically requires more iterations. Note that SuperCI,
DIIS and KDIIS should not be used when the active orbitals have an occupation of exactly 2.0 or 0.0! The
DIIS may sometimes converge slowly or trail towards the end such that real convergence is never reached.
The KDIIS [315] based on perturbation theory is an approximation to the regular DIIS procedure
avoiding redundant rotations. Both DIIS schemes avoid linear dependencies in the expansion space.
MaxDIIS 15 # max. no of DIIS vectors to keep

DIISThresh 1e-7 # overlap criteria for linear dependency
The combination of SuperCI and DIIS (switchstep) is particularly suited to protect the active space
composition. Adjusting the level shift will do the job. Level shift is the single most important lever
to control convergence.
# default = dynamic level-shifting based on Ext-Act, Int-Act

ShiftUp 2.0 # static up-shift the virtual orbitals
ShiftDn 2.0 # static down-shift the internal orbitals
MinShift 0.6 # minimum separation subspaces
Level-shift is particularly important if the active, inactive and virtual orbitals overlap in their orbital energies.
The energy separation of the subspaces is printed in the output. Ideally, the entries Ext-Act and Act-Int
should be positive and larger than 0.2 Eh. This will help the program to preserve your active space composition
throughout the iterations. If no shift is specified in the input, ORCA will choose a level-shift to guarantee an
energy separation between the subspaces (MinShift).
E(CAS)= -230.590325053 Eh DE= -0.000798832

In difficult cases the use of the Newton-Raphson method is highly recommended even if each individual
iteration is considerably more expensive. It is strong towards the end but it would be a waste to start
orbital optimization with the expensive NR method since its radius of quadratic convergence is quite small.
The computationally cheaper alternative is the SOSCF procedure belonging to the family of quasi-Newton
updates.
Keep in mind that the Newton-Raphson is designed for optimization on a convex surface (Hessian is
semidefinite). If the NR is switched on too early, there is a good chance that this condition is not fulfilled. In
this case the program will complain about negative eigenvalues or diagonal elements of the Hessian as can be
seen in the snippet below. The next optimization step is large and unpredictable. It is a wildcard that can
get you closer to convergence or immediate divergence of the CASSCF procedure.
||g|| = 0.771376945 Max(G)= 0.216712933 Rot=140,53

Warning: NEGATIVE diagonal element D(81,53)= -4.733590
Warning: NEGATIVE diagonal element D(82,53)= -4.737955
...
For larger system, the augmented Hessian equations are solved iteratively (NR iterations). The augmented
Hessian is considered solved when the residual norm, < r|r >, is small enough. Aside from the overall
CASSCF convergence, negative eigenvalues affect these NR iterations.

AugHess Tolerance (auto): Tol= 1.00e-07
AUGHESS-ITER 0: E= -0.174480747 <r|r>= 0.558679452
AUGHESS-ITER 1: E= -0.308672359 <r|r>= 0.468254671
AUGHESS-ITER 2: E= -0.434272813 <r|r>= 0.286305469
AUGHESS-ITER 3: E= -0.439149451 <r|r>= 0.286514628
AUGHESS-ITER 4: E= -0.605787445 <r|r>= 0.191691955
AUGHESS-ITER 5: E= -0.607766529 <r|r>= 0.310450670
AUGHESS-ITER 6: E= -0.611674930 <r|r>= 0.141402593
AUGHESS-ITER 7: E= -0.623145299 <r|r>= 0.394505306
AUGHESS-ITER 8: E= -0.658410333 <r|r>= 0.166915094
AUGHESS-ITER 9: E= -0.790571374 <r|r>= 4.722929453
AUGHESS-ITER 10: E= -0.790590554 <r|r>= 4.716012014
AugHess: No convergence in the Davidson procedure
...
There are a number of refined NR settings that influence the convergence behavior on a non-convex energy
surface. We mention the keywords for completeness and dis-encourage from changing the default settings.
If overall convergence cannot be changed due to negative eigenvalues, it is recommended to delay the NR
switchstep (switchconv, switchiter). This will require some trial and error, since the curvature of the
surface is a priori not know.
%casscf aughess
Solver 0 # Davidson (default)
1 # Pople (pure NR steps)
2 # DIIS
MaxIter 35 # max. no. of CI iters.
MaxDim 35 # Davidson expansion space
MaxDIIS 12 # max. number of DIIS vectors
UseSubMatrixGuess true # diag a submatrix of the Hessian
# as an initial guess
NGuessMat 512 # size of initial guess matrix (part of
# the Hessian exactly diagonalized)
ExactDiagSwitch 512 # up to this dimension the Hessian
# is exactly diagonalized (small problems)
PrintLevel 1 # amount of output during AH iterations
Tol 1e-6 # convergence tolerance
Compress true # use compressed storage
DiagShift 0.0 # shift of the diagonal elements of the
# Hessian
UseDiagPrec true # use the diagonal in updating
SecShift 1e-4 # shift the higher roots in the Davidson
# secular equations
UpdateShift 0.5 # shift of the denominator in the
# update of the AH coefficients
end
In general, convergence is strongly influenced by numerical noise, especially in the final iterations. One source
of numerical noise is the incremental Fock build. Thus, it can help to enforce more frequent full Fock matrix
formation.
ResetFreq 1 # reset frequency for direct SCF
If the orbital change in the active space is small, the active Fock matrix in ORCA is approximated using
the density matrix from the previous cycle saving a second Fock matrix build. However, this approximation
might also be a source of numerical instability. The threshold SwitchDens can be set to zero to enable
the exact build. The program default starts with a rather large value (1e-2) and will reduce this parameter
automatically when necessary.
switchdens 0.0001 # gtol * 0.1
In all of the implemented orbital optimization schemes the step-size correlates with the gradient-norm. With
the exception of the SuperCI PT, a constant damping factor can be set with the keyword GradScaling.
GradScaling 0.5 # constant damping in all steps

# Handling divergent steps (DE > 0.0)
There are situations when the active space has been chosen carefully, but the initial gradient is still far
off. To keep the good active space, we can suppress all rotation but the inactive-external ones until the
gradient-norm is small enough to continue safely. The threshold can be set with FreezeIE keyword. Once
the components of the gradient in the inactive-external direction have a weight of less than FreezeIE, all
constraints are lifted. ORCA by default freezes active rotations if the total gradient norm is larger than 1.0
and the active rotations have a weight of less than 5%. The feature can be turned off setting the threshold to
zero.
Similarly, rotations of the almost doubly occupied orbitals with the inactive orbitals can be damped using
the threshold FreezeActive. Rotations of this type are damped as long as all their weight is smaller than
FreezeActive. In contrast to the ShiftDn, it damps just the troublesome parts of internal-active rotations.
This option applies to all of the orbital optimization schemes but the SuperCI PT.
FreezeIE 0.4 # keep active space until int-ext rotation have

# a contribution of less than 40% to the ||g||
FreezeActive 0.03 # keep almost doubly occupied orbitals as long as
# their contribution is less than 3% to the ||g||
If the calculation starts from a converged Hartree-Fock orbitals, the core orbitals should not change dramatically
by the CASSCF optimization. Often trailing convergence is associated to rotations with low lying orbitals.
Their contribution to the total energy is fairly small. With the keyword FreezeGrad these rotations can be
omitted from the orbital optimization procedure.
FreezeGrad 0.2 # omit hitting a gradient norm ||g|| <0.2
The affected orbitals are printed at the startup of CASSCF.
FreezeGrad ... enabled if ||g|| is below 0.02

Note Convergence can be signaled if the reduced gradient reaches GTol
Last frozen orbital ... 9

First deleted orbital ... 320
Once rotations with core and deleted orbitals are stabilized they will be damped.
By default rotations with frozencore (or deleted virtuals) are not omitted. If the option FreezeGrad is active,
the ratio with respect to the total gradient is printed.
||g|| = 0.001240414 Max(G)= -0.000431747 Rot=319,1

--- Option=FreezeGrad: ||g|| = 0.001081707
= 87.21%
Omitting frozencore elements
Using the RI Approximation
A good way to speed up the calculations at the expense of only obtaining approximate results is to introduce
the RI approximation (works best in conjunction with a jk auxiliary bases). The RI approximation is
invoked by:
TrafoStep RI # RI used in transformation

# NOTE: NEEDS an Aux-Basis
Exact # exact transformation (default)
Monitoring the active space
During the iterations, the active orbitals are chosen to maximize the overlap with active orbitals
from the previous iterations. Maximizing the overlap does not make any restrictions on the nature of the
orbitals. Thus initially localized orbitals will stay localized and ordered, which is sometimes a desired feature
e.g. in the density matrix renormalization group approach (DMRG). This feature is set with the keyword
ActConstraints and is enabled by default (after the first 3 macroiterations). For some orbital optimization
procedures, such as the SuperCI, natural orbitals are more advantageous. Therefore, the ActConstraints
can be turned off in favor of natural orbital construction (see below). If the keyword is not set by the user,
ORCA picks the best choice for the given orbital optimization step.
ActConstraints 0 # no checks and no changes

1 # maximize overlap of active orbitals and check sanity. (default for DIIS)
2 # make natural orbitals in every iteration (default SuperCI)
3 # make canonical orbitals in every iteration
4 # localize orbitals
In addition to maximizing the overlap, "ActConstraints 1" checks if the composition of the active
space has changed i.e. an orbital has been rotated out of the active space. In this case, ORCA aborts and
stores the last valid set of orbitals. Below is an example error message.
--- Orbital Update [ DIIS]

--- Failed to constrain active orbitals due to rotations:
Rot( 37, 35) with OVL=0.960986
Rot( 38, 34) with OVL=0.842114
Rot( 43,104) with OVL=0.031938
In the snippet above, the active space ranges from 37-43. The program reports that orbitals 37,38 and 43
have changed their character. The overlap of orbital 43 (active) with the previous set of active orbitals is
just 3% and the program aborts. There are a number of reasons, why this happens in the calculation. If
this error occurs constantly with the same orbitals, it is worthwhile to inspect the rotating partner orbitals
(visualize). It might be sign that the active space is not balanced and should be extended. In many instances
changing the level-shift or lowering switchconv is sufficient to protect the active space. In some cases, turning
off the sanity check ("ActConstraints 0") and re-rotating orbitals will bring CASSCF closer to convergence.
Some problems can be avoided by a better design of the calculation. The CASSCF tutorial elaborates on the
subject in more detail.

There are situations such as parameter scans, where actconstraints is counter-productive and should be
disabled. In other words, we want to allow changes in the active space composition. As an example, consider
the rotation of ethylene around its double-bond represented by a CAS(2,2). Although the active space
consists of the bonding and anti-bonding orbitals -orbitals, their composition in terms of atomic orbitals
changes from the eclipsed to the staggered conformation. Depending on the actual input settings (orbstep
and number of scan points), this might trigger an abort.
Final orbitals options.
Once the calculation has converged, ORCA will do a final macro-iteration, where the orbital are finalized.
The default final orbitals are canonical in the internal and external space with respect to state-averaged
Fock operator. The active orbitals are chosen as natural orbitals. Other orbital types can be chosen for
the individual subspaces. Some of the options are specifically designed for coordination chemistry and are
discussed in the CASSCF tutorial that supplements the manual.
#internal space
IntOrbs CanonOrbs # canonical
LocOrbs # localized
unchanged # no changes
#external space
ExtOrbs CanonOrbs # canonical
LocOrbs # localized
DoubleShell # based on the shell and angular momentum
# of the highest active orbitals, the first few
# virtual orbitals correspond to the doubled-shell.
# All other virt. orbitals are canonicalized.
# For 3d-metal complexes, these are the 4d orbitals!
# For 4d-metal compleces, these are the 5d orbitals!
# And so on...
#active space
ActOrbs NatOrbs # natural
CanonOrbs # canonical
LocOrbs # localized
dOrbs # purify metal d-orbital and call the AILFT
fOrbs # purify metal f-orbital and call the AILFT
If the active space consists of a single set of metal d-orbitals, natural orbitals may be a mixture of the
d-orbitals. The active orbitals are remixed to obtain pure d-orbitals (ligand field orbitals) if the actorbs
is set to dorbs. The same holds for f-orbitals and the option forbs. Furthermore, the keyword dorbs
automatically triggers the ab initio ligand field analysis (AILFT). [348] The approach has been reported
in a number of applications. [349, 350] For more details on the AILFT approach, we refer to the respective
manual section and the CASSCF tutorial, where examples are shown. For a few applications, a printing of
the complete wavefunction is useful and can be requested.
PrintWF 0 # (default) prints only the CFGs

csf # Printing of the wavefunction in the basis of CSFs
det # Printing of the wavefunction in the basis of Determinants
The CI-step default setting is CSF based and is done in the present program by generating a partial formula
tape which is read in each CI iteration. The tape may become quite large beyond several hundred thousand
CSFs which limits the applicability of the CASSCF module. The accelerated CI (ACCCI) has the same
limitations, but uses a slightly different algorithm that handles multi-root calculations much more efficiently.
For now, properties (spin-orbit coupling, g-Tensor..) as well as NEVPT2 corrections are not available with
ACCCI. Nevertheless, it is the recommended option to converge a CASSCF calculation with multiple roots.
The resulting .gbw file may be used in a successive run to obtain properties or NEVPT2 corrections.
Larger active spaces are tractable with the DMRG approach or the iterative configuration expansion (ICE)
developed in our own group. DMRG and ICE return approximate full CI results. The maximum size of the
active space depends on the system and the required accuracy. Active spaces of 1020 orbitals should be
feasible with both approaches. The CASSCF tutorial covers examples with ACCCI and ICE as CI solvers.
%casscf
CIStep CSFCI # CSF based CI (default)
ACCCI # CSF based CI solver with faster algorithm for multi-root calculations
ICE # CSF based approximate CI -> ICE/CIPSI algorithm
DMRGCI # density matrix renormalization group approach instead of the CI
end
In the ICE approach, the computation of the coupling coefficients is time-consuming and by default repeated in
every macro-iteration. To avoid the reconstruction, it is recommended to once generate a coupling coefficient
library (cclib) and to use it for all of your ICE calculations. The details of the methodology and the cclib
are described in the ICE section 9.15.
Detailed settings for the conventional CI solvers (CISFCI, ACCCI, ICE) can be controlled in a sub-block.
Not all of the options and properties are available for CISteps apart from the default! NEVPT2,
transition densities and spin-dependent properties such as spin-orbit coupling are not yet available for ACCCI
and ICE.
%casscf ci
MaxIter 64 # max. no. of CI iters.
NGuessMat 512 # Initial guess matrix: 512x512
PrintLevel 3 # amount of output during CI iterations

ETol 1e-10 # default 0.1*ETol in CASSCF
RTol 1e-10 # default 0.1*ETol in CASSCF
TGen 1e-4 # ICE generator thresh
TVar 1e-11 # ICE selection thresh, default = TGen*1e-7
end
The CI-step DMRGCI interfaces to the BLOCK program developed in the group of G. K.-L. Chan [351354].
A detailed description of the BLOCK program, its input parameters, general information and examples
on the density matrix renormalization group (DMRG) approach, are available in the section 9.13 of the
manual.
The implementation of DMRG in BLOCK is fully spin-adapted. However, spin-densities and related
properties are not available in the current version of the BLOCK code. To start a DMRG calculation add
the keyword CIStep DMRGCI into a regular CASSCF input. ORCA will set default parameters and generate
and input for the BLOCK program. In general, DMRG is not invariant to rotation in the active space. The
program by default will run an automatic ordering procedure (Fiedler). More and refined options can be
set in the dmrg sub-block of CASSCF see section 9.13 for a complete list of keywords.
%casscf
nel 8
norb 6
mult 3
CIStep DMRGCI
# Detailed settings
dmrg
# more/refined options
...
end
end
It is highly recommended to start the calculation with split-localized orbitals. Any set of starting orbitals
(gbw file) can be localized using the orca loc program. Typing orca loc in the shell will return a small
help-file with details on how to setup an input for the localization. Examples for DMRG including the
localization are in the corresponding section of the manual 9.13. The utility program orca loc is documented
in section 9.32.5. Split-localization refers to an independent localization of the internal and virtual part of
the desired active orbitals.
NOTE:
Let us stress again: it is strongly recommended to first LOOK at your orbitals and make sure that the
ones that will enter the active space are really the ones that you want to be in the active space! Many
problems can be solved by thinking about the desired physical contents of the reference space before
starting a CASSCF. A poor choice of orbitals results in poor convergence or poor accuracy of the
results! Choosing the active orbitals always requires chemical and physical insight into the molecules
that you are studying!
Please try the program with default settings before playing with the more advanced options. If you
encounter convergence problems, have a look into your output, read the warning and see how the
gradient and energy evolves. Increasing MaxIter will not help in many cases.
Be careful with keywords such as !tightscf, !verytightscf and so on. These keywords set higher
integral thresholds, which is a good idea, but also tighten the CASSCF convergence thresholds. If you
do not need a tighter energy convergence, reset the criteria in the casscf block using ETol. More many
applications an energy convergence beyond 107 is unnecessary.
9.11.2 CASSCF Properties
The CASSCF program is able to calculate UV transition, CD spectra, SOC, Zeeman splittings, g-tensors,
magnetization, magnetic susceptibility and MCD spectra. The properties are exercised in more detail in
the CASSCF tutorial. The techniques used to calculate SOC, and Zeeman splittings are identical to those
implemented into the MRCI program. Input and keywords mimic the ones in the MRCI module described in
section 9.23.2. As an example, the input file to calculated g-values of CO+ is listed below:
!TZVPP Bohrs TightSCF #TightSCF for more accurate integrals

%casscf nel 9
norb 8
nroots 9
mult 2
switchstep NR
etol 1e-7 #reset energy convergence
rel
dosoc true #spin-orbit coupling (and ZFS)
gtensor true
end
end
* xyz 1 2
C 0 0 0.0
O 0 0 2.3504
*
Spin-Spin couplings are not accessible via the CASSCF module. However, it is possible to run CAS-CI type
calculations in MRCI. Converged CASSCF orbitals can be read setting the following flags
!MOREAD NOITER ALLOWRHF TZVPP TightSCF Bohrs

%moinp "convergedCASSCF.gbw"
%mrci
...
TPre 0.0
citype mrci
newblock 2 *
excitations none
refs CAS(9,8) end
end
soc
DoSSC true # spin-spin coupling
DoSOC true # spin-orbit coupling
...
end
end
* xyz 1 2
C 0 0 0.0
O 0 0 2.3504
*
A common way to introduce dynamical correlation for the property computation, is to replace the energies
entering the quasi-degenerate perturbation theory. If the NEVPT2 energy correction is computed in CASSCF,
there will be additional printings where CASSCF energies are replaced by the more accurate NEVPT2 values.
Alternatively, these diagonal energies can be taken from the input file similarly how it is described for the
MRCI module. A more detailed documentation is presented in the MRCI property section.
9.11.3 Fully Variational Spin-Orbit Coupled CASSCF
The fully variational spin-orbit coupled CASSCF approach allows the user to account for spin-orbit coupling
effects variationally. In this method the CASSCF wavefunction is presented as a linear combination of several
parts of pre-selected multiplicities S and all possible Ms = S, S 1, . . . , S.
REL = SMS + SMS 1 + . . . + S1MS 1 + S1MS 2 + . . . (9.176)
Each of the building block wavefucntions SMS is itself represented by a sum of CSFs
NX
CSF
SMS = ClSMS SM
l
S
(9.177)
l
The presented approach optimizes the CI coefficients and MO coefficients of a sum of the Born-Oppenheimer
(BO) Hamiltonian and the spin-orbit mean-field Hamiltonian (SOMF) that are given by
REL = H
H BO + H
SOMF (9.178)
The procedure employs the intermediate coupling scheme to account for SOC. This allows to employ
complex-valued CI coefficients, but real-valued molecular orbital coefficients.
As an example, the input file to calculated g-values of HgH is listed below:
! DKH-TZVP TightSCF Conv DKH Pmodel
%casscf
nel 3 # number of electrons
norb 5 # number of orbitals
nroots 3 # number of roots for the guess calculation
mult 2 # multiplicities
etol 1e-9
gtol 1e-05
maxiter 100
rel
dosoc true # accounts for the SOC interaction
soctype 1 # runs a variational calculation
0 # corresponds to the QDPT case
# NOTE: this is different from the soctype
# in the %rel block; see below
orbstep superci # SuperCI algorithm for convergence
diis # DIIS algorithm for convergence
qn # quasi-Newton algorithm for
# convergence
switchstep diis
gradscaling 0.1 # gradient scaling factor
maxiter 100 # number of iteration
0 # corresponds to the diagonalization
# of the BO and SOC operators in the
# basis of CSFs
printlevel 2
nroots 6 # number of relativistic roots
etol 1e-9
gtol 1e-3
socints 0 # SOC integrals are calculated once
# with guess MOs and used during
# every CASSCF iteration
1 # SOC integrals are re-calculated
# on every CASSCF calculation
-1 # SOC integrals are taken from files
# input name.mrci.sox.tmp etc
gtensor true # calculate g-tensor
usekramerssym true # use Kramers symmetry for the CI
# step
Weights = 0.166667, 0.166667, 0.166667, 0.166667, 0.166667, 0.166667
# weights for the state-averaging
ci
rtol 5e-14 # residual tolerance for the Davidson
# procedure
maxdim 10 # Davidson expansion space = MaxDim * NRoots
guesstype 0 # guess CI coefficient are obtained
# from diagonalization of
# a small CI matrix
1 # CI guess vectors a taken as final
# vectors from the previous CI
# iteration
end
end
end
%rel
soctype 2 # reads atomic densities from files
# 0.gbw for the atom 0, 1.gbw for the atom 1 etc.
4 # constructs atomic densities automatically
picturechange true # include picture change effects for
# for the SOC and the Zeeman operator
end
* xyz 0 2
Hg 0 0 0
H 0 0 1.766
*
Here, the non-relativistic (or scalar relativistic, real-valued) CI vectors are calculated first. They are used as
guess vectors for the following variational calculation. The relativistic procedure starts with the message
STARTING RELATIVISTIC CASSCF
INIT CI
SOLVE CI
ORBITAL-IMPROVEMENT-STEP:
Algorithm ... DIIS
Upward level shift ... 1.00e+00

Downward level shift ... 1.00e+00
AO integral handling ... CONVENTIONAL
Energy convergence tolerance ... 1.00e-09
Orbital gradient convergence ... 1.00e-03
Max. number of iterations ... 35
WEIGHTS: 0.166667 0.166667 0.166667 0.166667 0.166667 0.166667

First, the variational procedure diagonalizes the relativistic Hamiltonian in the basis of CSFs. This corresponds,
in fact, to the optimization of the CI coefficients using the Davidson procedure and guess MOs, and denoted
in the output as ZERO MACRO-ITERATION
ZERO MACRO-ITERATION
--------------------------------
COMPLEX DAVIDSON-DIAGONALIZATION
--------------------------------

Maximum number of CI iterations ... 64
Data storage type ... COMPLEX DOUBLE
Data compression type ... UNCOMPRESSED
At the end of the Davidson procedure the resulting energies and roots contributions are printed:
Eh cm-1 residual norm
-19596.632795883 0.00000 0.000000000000 # energies and

-19596.632795883 0.00001 0.000000000000 # residual norms
-19596.522641797 24176.02753 0.000000000000
-19596.522641797 24176.02754 0.000000000000
-19596.506502189 27718.26205 0.000000000000
-19596.506502189 27718.26205 0.000000000000
---------------------------------------------# relativistic
RELATIVISTIC CASSCF STATES # states are
---------------------------------------------# printed
ROOT 0: E= -19596.6327958833 Eh
S = 0.5 MS = 0.5 # contribution to

# the root 0 of CSFs
# with S = 1/2 and
# MS = 1/2
( -0.88880371, 0.17857507) 0 # real and image parts

# of the CI
# coefficient of the
# CFS "0"
( -0.02017328, 0.00746783) 1
( -0.00746783, -0.02017328) 2
( -0.08345160, 0.01676678) 3
( -0.02929834, 0.00588651) 4
( -0.08258516, 0.01659270) 9
( 0.07876099, -0.01582436) 10
( 0.04482426, -0.00900592) 11
( 0.04482426, -0.00900592) 16
( -0.02273506, 0.00456784) 19
( 0.05706131, -0.01146454) 22
( 0.03892008, -0.00781967) 23
( 0.03892008, -0.00781967) 28
( 0.05898014, -0.01185007) 31
S = 0.5 MS = -0.5
( 0.09374074, 0.34621370) 0
( -0.00586884, -0.05405190) 1
( -0.05405190, 0.00586884) 2
( 0.00880151, 0.03250671) 3
( 0.00309005, 0.01141252) 4
( 0.00871013, 0.03216921) 9
( -0.00830680, -0.03067959) 10
( -0.00472754, -0.01746029) 11
( -0.00472754, -0.01746029) 16
( -0.00601817, -0.02222696) 22
( -0.00410484, -0.01516045) 23
( -0.00410484, -0.01516045) 28
( -0.00622054, -0.02297440) 31
Next, the properties specified in the input are calculated. Next, the obtained complex CI coefficients are
used to develop the orbital gradient as it is described in detail in the literature [355]. The orbital gradient is
used for optimizing real-valued MOs.
MACRO-ITERATION 1:
Scaling of the SOC contribution 1.000
--------------------------------
COMPLEX DAVIDSON-DIAGONALIZATION
--------------------------------

Maximum number of CI iterations ... 64
Data storage type ... COMPLEX DOUBLE
Data compression type ... UNCOMPRESSED
After the CASSCF calculation converged
-19596.632736469 0.00000 0.000000000000 ( 0.00)

-19596.632736469 0.00000 0.000000000000
-19596.522536003 24186.20658 0.000000000000
-19596.522536003 24186.20659 0.000000000000
-19596.506572822 27689.71986 0.000000000000
-19596.506572822 27689.71986 0.000000000000

E(CAS)=-19596.553948432 Eh DE= 0.000000510
CI-PROBLEM SOLVED
Making density matrices
SOC Part:
||g||= 0.024945604 Max(G)= -0.012476958 Rot=111,43
||g||= 0.000827087 Max(G)= -0.000742821 Rot=111,43
---- THE CAS-SCF GRADIENT HAS CONVERGED ---
the complex CI vectors are printed again using the same format as indicated above. The program is able to
calculate UV transitions, SOC splittings, and g-tensors for doublet states. The options are similar to those
used for the CASSCF used to calculate SOC splittings are identical to those implemented into the MRCI
program. The output for the calculated g-tensor of HgH is printed in the same format as produced the by
MRCI program.
-------------------
KRAMERS PAIR 1 :
-------------------
Matrix elements Re<1|S|1> -0.441689 -0.391275 0.793281
Matrix elements Re<1|S|2> 0.272168 -0.908540 -0.292499
Matrix elements Im<1|S|2> -0.846822 -0.087922 -0.507771
Matrix elements Re<1|L|1> 0.039367 0.034874 0.005531
Matrix elements Re<1|L|2> -0.024258 0.080977 -0.002039
Matrix elements Im<1|L|2> 0.075476 0.007836 -0.003540
-------------------
ELECTRONIC G-MATRIX
-------------------
raw-matrix g-matrix:
0.496395 -1.657046 -0.589760
1.544481 0.160356 1.023809
-0.805576 -0.713629 1.599477
-------------------
ELECTRONIC gTg-MATRIX
-------------------
3.280781 -0.000000 0.000000

-0.000000 3.280781 0.000000
0.000000 0.000000 3.954328
diagonalized g**2 matrix (sqrt of eigenvalues taken):
1.811293 1.811293 1.988549
0.993455 0.114223 0.000000

0.114223 -0.993455 0.000000
-0.000000 0.000000 1.000000
9.12 N-Electron Valence State Pertubation Theory 427
g-Shifts in ppm (for completeness; 1e6*(g[i]-gel))

g1 = -191026.7
g2 = -191026.7
g3 = -13770.1
9.12 N-Electron Valence State Pertubation Theory
CASPT2 and NEVPT2 belongs to the family of internally contracted perturbation theories with CASCI
reference wavefunctions. Several studies indicate that CASPT2 and both variants of NEVPT2 produce
energies of similar quality. [119, 120] The NEVPT2 methodology developed by Angeli et al exists in two
formulations namely the strongly-contracted NEVPT2 (SC-NEVPT2) and the partially contracted NEVPT2
(PC-NEVPT2). [116118] Irrespective of the name partially contracted coined by Angeli et al, the latter
approach employs a fully internally contracted wavefunction (FIC). Hence, we use the term FIC-NEVPT2
in place of PC-NEVPT2. ORCA features both variants. Fully internally contracted and strongly contracted
NEVPT2 differ in the basis of the wavefunction expansion. SC-NEVPT2 employs strongly contracted CSFs,
which form a compact and orthogonal basis making it computationally slightly more attractive. Hence, the
SC-NEVPT2 has been the program default for long time. NEVPT2 has many desirable properties - among
them:
It is intruder state free due to the choice of the Dyall Hamiltonian [347] as the 0th order Hamiltonian.
The 0th order Hamiltonian is diagonal in the perturber space. Therefore no linear equation
system needs to be solved.
It is strictly size consistent. The total energy of two non-interacting systems is equal to the sum of
two isolated systems.
It is invariant under unitary transformations within the active subspaces.
strongly contracted: Perturber functions only interact via their active part. Different subspaces
are orthogonal and hence no time is wasted on orthogonalization issues.
fully internally contracted: Invariant to rotations of the inactive and virtual subspaces.
The principal methods are called within the CASSCF block and detailed settings can be adjusted in the
nevpt subblock. We will go through some of the detailed setting in the next few subsections. For historical
reasons, some features, such as the quasi-degenerate NEVPT2, are only available for the strongly contracted
NEVPT2.
Note that methodology by default employs the frozencore approximation, which can be disabled with the
simple keyword !NoFrozenCore. Frozencore calculations are not significantly faster for the trade-
off in accuracy and hence not recommended. Nevertheless, it is the default to be consistent with the
rest of the ORCA program package. A complete description of the frozecore settings can be found in section
9.8.
%casscf
...
CIStep DMRGCI # optional to run DMRG-NEVPT2
# default: CSFCI
trafostep ri # RI approximation for CASSCF and NEVPT2
nevpt2 SC # strongly contracted NEVPT2
FIC # fully internally contracted / partially contracted NEVPT2
DLPNO # FIC-NEVPT2 using the DLPNO framework for large molecules
# detailed settings (reduced list). This is optional!
nevpt
NThresh 1e-6 # FIC-NEVPT2 cut off for linear dependencies
D4Tpre 1e-10 # Truncation of the 4-pdm
D3Tpre 1e-14 # Trunaction of the 3-pdm
canonstep 1 # default (exact):canonical orbitals for each state
QDType 1 # QD-SCNEVPT2, default = 0
EWIN -3,1000 # Energy window for the frozencore setting fc_ewin
D4Step Fly # 4-pdm is constructed on the fly
F12 true # F12-Correction
end
end
NEVPT2 can also be set using the simple keywords on top of any valid CASSCF input.
!SC-NEVPT2 # for the strongly contracted NEVPT2

!FIC-NEVPT2 # for the fully internally contracted NEVPT2
!DLPNO-NEVPT2 # for the DLPNO variant of the FIC-NEVPT2
! ...
%casscf ...
The two computationally most demanding steps of the NEVPT2 calculation are the initial integral transfor-
mation involving the two-external labels and the formation of the fourth order density matrix (D4). Efficient
approximations to both issues are available in ORCA.
If not otherwise specified (keyword CIStep), CASSCF and consequently NEVPT2 use a conventional CSF
based solver for the CAS-CI problem. The NEVPT2 methodology can be combined with approximate CI
solution such as the DMRG approach described in section 9.13. With ORCA 4.0 NEVPT2-DMRG calculations
for the FIC and SC type wavefunction are available using the methodology developed by the Chan group. [356]
Aside from the usual DMRG input, the program requires an additional parameter (nevpt2 MaxM) in the
DMRG block.
cistep DMRGCI
%dmrg
...
nevpt2_MaxM 2000 # see Guo, Chan et al. [356]

end
NEVPT2 SC # or FIC (numerical more sensitive)
In this case, the exact fourth-order density matrix is stored on disk. Using the RI approximation, large
molecules with actives spaces of up to 20 orbitals should be computable. The DMRG extension with DLPNO
and F12 are evaluated and locked until the next release. Future version might also support the CIStep ACCCI
and CIStep ICE.
RI, RIJK and RIJCOSX Approximation
Setting the RI approximation on the CASSCF level, will set the RI options for NEVPT2 respectively. The
three index integrals are computed and partially stored on disk. Three index integral with two internal labels
are kept in main memory. The two-electron integrals are assembled on the fly.
%casscf
...
trafostep ri # enable RI approximation in CASSCF and NEVPT2
nevpt2 SC # or the NEVPT2 approach or your choice
end
Additional speedups can be obtained if the Fock operator formation is approximated using the !RIJCOSX or
!RIJK techniques. Note that NEVPT2 and CASSCF program parts accept a single auxiliary basis. Aside
from the Fock operator, the auxiliary basis is required for the integrals appearing in the CASSCF orbital
gradient and in the NEVPT2 correlation energy. Hence, a sufficiently large and balanced auxiliary basis
should be chosen.
#RIJCOSX one-liner
! def2-svp def2/jk RIJCOSX
#RIJK one-liner, conv is mandatory for RIJK in CASSCF
! def2-svp def2/jk RIJK conv
The described methodology allows the computation of systems with up to 2000 basis functions. Even larger
molecules are accessible in the framework of DLPNO-NEVPT2 described in the next subsection. Several
examples can be found in the CASSCF tutorial.
Beyond the RI approximation: DLPNO-NEVPT2
For systems with more than 80 atoms, we recommend the recently developed DLPNO-NEVPT2. [121] It
is a successful combination of DLPNO strategy with the FIC-NEVPT2 method. As its single reference
counterparts, DLPNO-NEVPT2 recovers 99.9% of the FIC-NEVPT2 correlation energies even for large
system. The input structure is similar to the parenting FIC-NEVPT2 method. Below you find an input
example for the Fe(II)-complex depicted in 9.4, where the active space consists of the metal-3d orbitals.
The example takes about 9 hour (including 3 hour for one CASSCF iteration) using 8 cores (2.60GHz Intel
E5-2670 CPU) for the calculation to finish. A detailed description of the DLPNO-NEVPT2 methodology can
be found in our article. [121].
# DLPNO-NEVPT2 calculation for quintet state of FeC72N2H100

!PAL8 def2-TZVP def2/jk
!moread noiter
%moinp "FeC72N2H100.gbw-CASSCF"
%MaxCore 8000
%casscf
nel 6
norb 5
mult 5
trafostep ri # ri approximation is mandatory for DLPNO-NEVPT2
nevpt2 DLPNO
# sub-block with optional settings to control the accuracy
nevpt
TCutPNO 1e-8 # Most important parameter controlling the accuracy (default 1e-8)
NThresh 1e-6 # FIC-NEVPT2 cut off for linear dependencies (default 1e-6)
end
end
*xyz 0 5 FeC72N2H100.xyz
Just like RI-NEVPT2, the calculations requires an auxiliary basis. The aux-basis should be of /c or /jk type
(more accurate). Aside from the paper of Yang et al, a concise report of the accuracy can be found in the
CASSCF tutorial, where we compute exchange coupling parameters. Note that in the snippet above, we have
repeated some of the default setting in the NEVPT sub-block. This is not mandatory and should be avoided
to keep the input as simple as possible.
As mentioned earlier, the CASSCF step can be accelerated with the RI JK or RIJCOSX approximation.
Both options are equally valid for the DLPNO-NEVPT2. The RIJK variant typically produces more accurate
results than RIJCOSX. The input file is almost the same as before, except for the keyword line:
# The combination of RIJK with DLPNO-NEVPT2

!PAL8 def2-TZVP def2/JK conv RIJK
Figure 9.4: Structure of the FeC72 N2 H100
Explicitely correlated NEVPT2: NEVPT2-F12
Like single-reference MP2, the NEVPT2 correlation energy converges slowly with the basis set. Aside from
basis set extrapolation, the R12/F12 method are popular methods to reach the basis set limit. For comparison
of F12 and extrapolation techniques we refer to the study of Liakos et al. [84] ORCA features an F12
correction for the FIC-NEVPT2 wavefunction using the RI approximation. [357] The RI approximation is
mandatory as the involved integrals are expensive. Just like in MP2-F12, the input requires an F12 basis, an
F12-cabs basis and a sufficient RI basis (/jk or /c).
# aug-cc-pvdz/C used as RI basis

! cc-pvdz-F12 aug-cc-pvdz/C cc-pvdz-f12-cabs
%casscf nel 8
norb 6
mult 3,1
gtol 1e-6
etol 1e-14
nevpt2 2 #FIC-NEVPT2
trafostep ri #RI approximation must be on
nevpt
F12 true #Do the F12 correction

end
end
*xyz 0 3
O 0.0 0.0 0.0
O 0.0 0.0 1.207
*
Approximations for large active CASSCF space
For CASSCF spaces of (8,8) and larger the formation of the fourth order density matrix becomes more and
more the time dominating step of the NEVPT2 calculation. To improve this situation, ORCA truncates
the CASSCF wavefunction during the formation of the fourth and third order density matrices. Only
configurations with a weight larger than a given parameter d4tpre are taken into account. The same
approximation is available for the third order density matrix controlling d3tpre. Both of the parameters can
be adjusted within the nevpt of casscf.
%casscf
...
nevpt2 SC # or whatever NEVPT2 approach or your choice
nevpt # NEVPT sub-block for NEVPT2 specific settings

d4tpre 1e-10 #default
d3tpre 1e-14 #default
end
end
These approximations naturally can affect the configuration RI as well. In this context, it should be
noted that a configuration corresponds to a set of configuration state functions (CSF) with identical orbital
occupation. For each state the dimension of the CI and and RI space is printed.
D3 Build ... CI space truncated: 141 -> 82 CFGs

... RI space truncated: 141 -> 141 CFGs
D4 Build ... CI space truncated: 141 -> 82 CFGs
... RI space truncated: 141 -> 141 CFGs
The default values usually produce errors of less than 1 mEh. However, the error introduced by the d4tpre is
system dependent and should be double checked. The exact NEVPT2 energy is recovered with the parameters
set to zero. The approximation is available for all variants of NEVPT2 (SC, FIC and DPLNO-FIC).
Storage of the fourth order density matrix can easily reach several gigabytes and thus cannot be kept in core
memory for a large active space. We have implemented a few strategies for the computation of fourth order
density matrix:
1. fly , compute elements and immediately contract them
2. lfly, less memory demanding version of fly
3. on disk, the full density matrix is stored on disk
4. in core, the full density matrix is kept in core memory
%casscf
...
d4step disk # dumped on disk
fly # on the fly (default)
core # in core memory
lfly # less memory demanding
# compared to fly, but substantially slower!
end
end
State-averaged NEVPT2
In the definition of the Dyall Hamiltonian [347] the CASSCF orbitals are chosen to diagonalize the Fock
operator (pseudo-canonicalized). Therefore, using a state-averaged CASSCF wave function, the NEVPT2
procedure involves the construction and diagonalization of the state-specific Fock operators and is thus
resulting in a unique set of orbitals for each state. This becomes quickly inefficient for large number of states
or large molecular systems since each orbital set implies an integral-transformation. This is the default
setting for NEVPT2 and is printed in the output
NEVPT2-SETTINGS:
Orbitals ... canonical for each state
Other orbital options can be set using the keyword canonstep.
%casscf
...
canonstep 0 # state-averaged orbitals and specific orbital energies
1 # canonical for each state
2 # state-averaged orbitals and orbital energies
3 # 1-step orbital relaxation
# and canonical for each state (partially relaxed)
end
end
The final orbitals of the state-averaged CASSCF diagonalize the state-averaged Fock operator. Large
computational savings can be made if these orbitals are employed for all of the states. canonstep 0 chooses
orbital energies as diagonal elements of the state-specific Fock operators. In previous version of ORCA,
this has been the default setting. These options work best if the averaged states are similar in nature. For
SC-NEVPT2, we have implemented two more canonsteps, which trade accuracy for speed and vice versa.
canonstep 2 is more approximate and employs orbital energies from the state-averaged calculation. Thus
there is no contribution to excitation energies from the perturber class Vijab at this level of approximation.
If the states under consideration are substantially different, these approximations will be of poor quality and
should be turned off. Better results can be achieved, if the state-averaged orbitals are partially relaxed for
each state before the actual SC-NEVPT2 calculation. [358] Often it is not possible to optimize the excited
states separately. Thus canonstep 3 will try a single steepest descent step for each state before running the
actual SC-NEVPT2 calculation with canonicalized orbitals. Optionally, instead of a steepest descent using
an approximate diagonal Hessian, a single Newton-Raphson step can be made.
%casscf
...
NEVPT2 SC
gstep SOSCF true # steepest descent step
NR false # Newton-Raphson step
end
end
Despite a converged state-averaged calculation, the gradient for the individual states can be surprisingly large.
As a consequence, the orbital relaxation might fail as both methods might be outside their convergence radius.
ORCA will retry the relaxation with an increased damping. If the orbital update still fails, the program will
stick with the initial orbitals. Setting an overall damping manually, might help the relaxation procedure.
NEVPT2 SC
nevpt
gscaling 0.5 # damp gradient with a pre-factor
end
Quasi-Degenerate NEVPT2
NEVPT2 as it is presented in the previous subsections follows the recipe of diagonalize and perturb. The
0th order wavefunction is determined by the diagonalization of the CAS-CI matrix. The space spanned by
the CAS-CI vectors is often referred to as model space. The subsequent perturbation theory is constructed
based on the assumption that the states under consideration are well described within the model space.
(1)
Consequently, the first order correction to the wavefunction I does not affect the composition of the
reference state |Ii. Corrections to the wavefunction and energy arise from the interaction of the reference
state with the functions |ki of the contributing first order interacting space
(1)
X
I = Ck |ki (9.179)
k
(2)
X hI |H| ki hk |H| Ii
EI = (0)
(9.180)
k EI Ek
This is problematic, when the interaction/mixing of states are falsely described at the CASSCF level. A
typical example is the dissociation of lithium fluoride.
!def2-tzvp nevpt2 nofrozencore

%casscf
nel 2
norb 2 #Li(2s), F(2pz)
mult 1
nroots 2
end
%paras
r = 3,7,200
end
*xyz 0 1
Li 0 0 0
F 0 0 {r}
*
Here, the ground and first excited state of + should not cross. However, at the NEVPT2 level, an erratic
double crossing is observed instead.
-106,99
-107
-107,01
Emergy [Eh]
-107,02
-107,03
QDNEVPT2
-107,04
NEVPT2
-107,05
-107,06
3 3,5 4 4,5 5 5,5 6 6,5 7
Li-F [Angstrom]
Figure 9.5: NEVPT2 and QDNEVPT2 Li-F dissociation curves of the ground and first excited
states for a CAS(2,2) reference
A re-organizing of the reference states can be introduced in the framework of quasi-degenerate perturbation
theory. In practice, an eff. Hamiltonian is constructed allowing off-diagonal corrections to the second order
energy
(0)
X hI |H| ki hk |H| Ji
HIJ = IJ EI + (0)
(9.181)
k EI Ek
Diagonalization of this eff. Hamiltonian yields improved energies and rotation matrix (right eigenvectors) that
introduces the desired re-mixing of the reference states. The quasi-degenerate extension to SC-NEVPT2 [359]
can be switched on with the keyword qdtype.
%casscf
...
nevpt2 SC
nevpt
qdtype 0 # no QDPT2
1 # non-Hermetian eff. Hamiltonian
2 # Hermitian eff. Hamiltonian (Cloizeaux)
end
end
ORCA will print the eff. Hamiltonian matrix and its eigenvectors at the end of the calculation.
===============================================================
QD-NEVPT2 Results
===============================================================
*********************
MULT 1,
*********************
Total Hamiltonian to be diagonalized

0 1
0 -107.074594 -0.012574
1 -0.011748-107.003810
Right Eigenvectors
0 1
0 -0.987232 0.170171
1 -0.159292 -0.985414
--------------------------
ROOT = 0
--------------------------
--------------------------
ROOT = 1
--------------------------
By construction the Hamiltonian is non-Hermitian. If the matrix is ill-conditioned, the eigenvalues can
be imaginary. ORCA prints a warning in this case. The energy denominator in the quasi-degenerate
NEVPT2 is very sensitive to approximations. The canonicalization options with averaged orbitals and
orbitals energies (canonstep 0/2) have the tendency to lessen the energy-denominator. To avoid artifacts,
the calculation is restricted to canonstep 1 each state has its own orbitals. For the computation of
properties e.g. expectation values, the non-Hermitian formulation requires left- and right eigenvectors. A
Hermitian variant [360] with a single set of eigenvectors (right) can be constructed. The transformation
does not change the energies but affects the mixing of states. Despite the simplification, the recommended
option is the non-Hermitian variant!
If properties are requested within the casscf module i.e. zero-field splitting, there will be an additional printing
with the improved CI vectors and energies. For technical reasons, properties that are not computed in
CASSCF such as the M ossbauer parameters do not benefit from the QD-NEVPT2 correction.
9.13 Density Matrix Renormalization Group
The BLOCK code in ORCA is only available on the Linux platform!
BLOCK is an implementation of the density matrix renormalization group (DMRG) algorithm from the Chan
group. [351354, 361] The references given should be cited when using this part of the program.
The DMRG is a variational wavefunction method. It can be viewed as (i) an efficient method for strong
correlation in large complete active spaces, (ii) a brute force method to systematically approach FCI for a large
number of electrons and orbitals, (iii) a polynomial cost route to exact correlation in pseudo-one-dimensional
molecules, such as chains and rings.
Although the algorithm is somewhat complicated compared to many quantum chemistry methods, significant
effort has been devoted in BLOCK to ensure that it can be run in a simple black-box fashion. In most cases,
only a single keyword needs to be specified.
To provide an idea of how the DMRG can be used, here are some examples. The timings will vary depending
on your computational setup, but the following are calculations that run in a few hours to a day, on a single
12-core Xeon Westmere cluster node:
Complete active space (CAS) CI calculations for active spaces with up to 30 electrons in 30 active
orbitals, targetting up to 110 states, e.g. Jacobsens catalyst in a 32 electron, 25 orbital active space,
One-dimensional chain molecules, with widths of up to 4 orbitals, and about 100 orbitals in total,
e.g. the -active space of a 425 graphene nanoribbon,
FCI benchmark solutions in molecules with fewer than 20 electrons, and up to 100 orbitals, e.g. C2 in
a cc-pVTZ basis, D2h symmetry (12 electrons in 60 orbitals),
Accuracies in energy differences or total energies of about 1 kcal/mol.
The following are calculations which are possible with the BLOCK code, but which are challenging, and require
large memory (e.g. up to 8 GB per core) and computational time (e.g. from a day to more than a week on
up to 6 12-core Xeon Westmere nodes),
Complete active space (CAS) CI calculations in active spaces with around 40 electrons in 40 active
orbitals, targetting a few states, for example, an Fe(II)-porphine (40 electrons in 38 orbitals) with
an active space of Fe 3d, 4d and all porphine and donor orbitals, or an Fe 3d, S 3p active space
calculation for [Fe4 S4 (SCH3 )4 ]2 ,
One-dimensional chain molecules, with widths of up to 6 orbitals, and about 100 total orbitals,
Champion FCI benchmark solutions in small molecules, such as butadiene in a cc-pVDZ basis (22
electrons in 82 orbitals),
Accuracies in energy differences or total energies of about 1 kcal/mol.
If any these calculations interest you, then you might want to try a DMRG calculation with BLOCK!
9.13 Density Matrix Renormalization Group 439
9.13.1 Technical capabilities
Currently, BLOCK implements the following
An efficient DMRG algorithm for quantum chemistry Hamiltonians
Full spin-adaptation (SU(2) symmetry) and Abelian point-group symmetries
State-averaged excited states
Note that the standalone version of BLOCK may provide more capabilities than are available through the
external interface. See the BLOCK website for details [362].
9.13.2 How to cite
We would appreciate if you cite the following papers in publications resulting from the use of BLOCK :
G. K.-L. Chan and M. Head-Gordon, J. Chem. Phys. 116, 4462 (2002),
G. K.-L. Chan, J. Chem. Phys. 120, 3172 (2004),
D. Ghosh, J. Hachmann, T. Yanai, and G. K.-L. Chan, J. Chem. Phys., 128, 144117 (2008),
S. Sharma and G. K-.L. Chan, J. Chem. Phys. 136, 124121 (2012).
In addition, useful DMRG references relevant to quantum chemistry can be found in the review below by
Chan and Sharma.
G. K-.L. Chan and S. Sharma, Ann. Rev. Phys. Chem. 62, 465 (2011),
9.13.3 Overview of BLOCK input and calculations
Within ORCA, the BLOCK program is accessed as part of the CASSCF module. BLOCK can be run in two
modes: CASCI mode (no orbital optimization) or CASSCF mode. To enable CASCI mode, set maxiter
1.
%casscf
maxiter 1 # remove if doing CASSCF
CIStep DMRG
...
end
For small molecule CASCI it may be possible to correlate all orbitals. In general, similar to a standard
CASSCF calculation, it is necessary to select a sensible active space to correlate. (See Section 9.13.4.1 on
CASSCF). This is the responsibility of the user.
9.13.4 Standard commands
Once the orbitals to correlate have been chosen, and the wavefunction symmetries and quantum numbers are
specified, the accuracy of the DMRG calculation is governed by two parameters: the maximum number of
renormalized states M ; and, the order and localization of the orbitals.
The most important parameter in the DMRG calculation is M , the number of renormalized states. This
defines the maximum size of the wave-function expansion, which is O(M 2 ) in length in the renormalized
basis. As M is increased, the DMRG energy converges to the exact (FCI or CASCI) limit.
The DMRG maps orbitals onto a 1D lattice, thus the best results are achieved if strongly interacting orbitals
are placed next to each other. For this reason, the DMRG energy is not generally invariant to orbital rotations
within the active space, and orbital rotation and ordering can improve the DMRG energy for a given M . As
M is increased, the DMRG energy becomes less and less sensitive to the orbital ordering and localization.
To minimize the number of wavefunction optimization steps, it is often advantageous to perform DMRG
calculations at small M , then increase M to the final maximum value. This sequence of optimizations is
governed by the sweep schedule, which specifies how many optimization steps (sweeps) to perform at each
intermediate value of M .
The above may seem to make running a DMRG calculation more complicated than a usual quantum chemistry
calculation, however, BLOCK provides a set of default settings which eliminate the need to specify the above
parameters by hand. We highly recommend that you first learn to use the BLOCK program with these default
settings. In the default mode, the orbitals are ordered automatically (Fiedler vector method [363366]) and a
default sweep schedule is set.
An example of a default CASCI calculation on the C2 molecule correlating all electrons in a VTZ basis, is
given here:
!cc-pvtz pal4
%MaxCore 16000
%casscf
nel 8
norb 58
nroots 1
mult 1
maxiter 1
CIStep DMRGCI
DMRG
maxM 5000
end
end
* xyz 0 1
C 0 0 -0.621265
C 0 0 0.621265
*
Once you are familiar with the default mode, we recommend exploring the localization of orbitals. In general,
DMRG benefits from the use of localized orbitals, and these should be used unless the high-symmetry of
the molecule (e.g., D2h symmetry) provides compensating computational benefits. We recommend using
split-localized orbitals, which correspond to localizing the occupied and virtual orbitals separately. An
example of a split-localized default DMRG calculation on the porphine molecule, correlating the full -space
(26 electrons in 24 orbitals), in a cc-pVDZ basis is given in Sec. 9.13.5.
For a given maxM, it can take a long time to tightly converge DMRG calculations (e.g. to the default 1e-9
tolerance). To decrease computation time, you may wish to loosen the default tight sweep tolerance or control
the maximum number of sweep iterations with the commands sweeptol and maxIter.
9.13.4.1 Orbital optimization
Orbital optimization (mixing the external/internal space with the active space, not to be confused with
orbital rotation and ordering in the active space) in DMRG calculation can be performed by using the BLOCK
program as the CIStep within a CASSCF calculation, as described above. For the moment, spin-densities
and related properties are not available for this CIStep.
During the optimization iterations it is important that the active orbitals maintain their overlap and ordering
with previous iterations. This is done using actConstrains. This flag is set by default.
%casscf
ActConstrains 1 # maintain shape and ordering of active orbitals
...
end
In general, performing a DMRG calculation with orbital optimization is quite expensive. Therefore, it is often
best to carry out the orbital optimization using a small value of maxM (enabled by the default parameters
maxM=25 and the resulting sweep schedule), and to carry out a final single-point calculation using a larger
value of maxM.
9.13.4.2 Advanced options
There may be times when one wants finer control of the DMRG calculation. All keywords are shown in the
complete set of BLOCK options 9.13.4.4 below. The startM command allows to change the starting number
of states in DMRG calculations. It is also possible to specify the entire sweep schedule manually. A sweep
schedule example follows:
%casscf
...
dmrg
MaxIter 14
switch_rst 1e-3
TwoDot_to_oneDot 12
NSchedule 3
sche_iteration 0, 4, 8
sche_M 50, 100, 500
sche_sweeptol 1e-4, 1e-6, 1e-9
sche_noise 1e-8, 1e-11, 0.0
end
end
The commands above are:
MaxIter, corresponds to the maximum number of sweeps done by DMRG;
NSchedule, specifies the total number of schedule parameters we will specify;
Sche iteration, details the sweep number at which to change the parameters of the calculation.
Notice count begins at 0;
Sche M, is the number of renormalized states at each sweep;
Sche sweeptol, is the tolerance of the Davidson algorithm;
Sche noise, is the amount of perturbative noise we add each sweep;
Twodot to onedot, specifies the sweep at which the switch is made from a twodot to a onedot algorithm.
The recommended choice is to start with twodot algorithm and then switch to onedot algorithm a few
sweeps after the maximum M has been reached. To do a calculation entirely with the twodot or the
onedot algorithm, replace the twodot to onedot line with twodot 1 or onedot 1;
switch rst, defines the switching threshold of orbital gradient below which DMRG turns to onedot
algorithm and restarts from previous operators and wavefunction. This is essential to avoid oscillation
of energy values in the orbital optimization.
The default DMRG sweep schedule is selected automatically according to the choice of computational mode.
By default two different sets of predefined schedules are supported for CASCI and CASSCF computations,
respectively.
In CASCI mode, the default schedule corresponds to the following: starting from a given startM (where the
default is 250 and 8 sweeps), increase to a value of 1000 (8 sweeps) and increment by 1000 every 4 iterations
until maxM is reached. The algorithm switches from twodot to onedot two sweeps after the maxM has been
reached.
In CASSCF mode, the orbital optimization requires much fewer renormalized states to converge the wavefunc-
tion with respect to orbital rotations. The default schedule therefore starts with startM (where the default
is 25 and 2 sweeps), and increments by a factor of 2 every 2 sweeps util maxM is reached. The algorithm
continues the sweep at maxM by decreasing the Davison tolerance sche sweeptol and noise level sche noise
every 2 cycles by a factor of 10, until sche sweeptol becomes smaller than sweeptol.
For better control of the orbital ordering, we also provide a genetic algorithm minimization method of a
weighted exchange matrix. The genetic algorithm usually provides a superior orbital ordering to the default
ordering, but can itself take some time to run for large numbers of orbitals. The genetic algorithm can be
enabled by
%casscf
...
DMRG
auto_ordering GAOPT
end
end
within the %casscf input.
9.13.4.3 Troubleshooting
The two most common problems with DMRG calculations are that (i) convergence with maxM is slower than
desired, or (ii) the DMRG sweeps get stuck in a local minimum. (i) is governed by the orbital ordering /
choice of orbitals. To improve convergence, turn on the genetic algorithm orbital ordering.
If you suspect (ii) is occurring, the simplest thing to do is to increase the starting number of states with the
startM (e.g. from 500 to 1000 states). Local minima can also sometimes be avoided by increasing the noise
in the DMRG schedule, e.g. by a factor of 10. To check that you are stuck in a local minimum, you can carry
out a DMRG extrapolation (see extended Manual in the BLOCK website).
Note that the present DMRG-SCF establishes the input order of active space orbitals according to their
Hartree-Fock occupancy, even if these orbitals are ultimately canonical or split-localized canonical in nature.
This is specified by hf occ in which the Hartree-Fock occupancy is derived by default from the one-electron
integrals. Other options for obtaining the occupancy are available (see 9.13.4.4).
Somet times the energy values produced from one SCF cycle to another may oscillate. Such a nonlinear
numerical behaviour may occur typically by the last few iterations, most likely caused by the loss of a
certain distribution of quantum numbers (eg, particle number, irrep symmetry and spin) in the blocking and
decimation procedure due to incomplete many-body basis. On the other hand, the loss of quantum numbers
is the main source of energy discontinuities on potential energy curves calculated by DMRG-SCF using a
small number of renormalized states.
In the current release of DMRG-SCF implementation, the number of quantum states is locked to avoid these
problems. The locking mechanism is turned on when the orbital gradient falls below a certain threshold
defined by the keyword switch rst (default: 0.001). The DMRG calculation then starts from previous
operators and wavefunction in which a perturbative noise is not added. Locking quantum states and restaring
DMRG wavefunction not only ensures a smooth convergence towards the final energy but also minimizes the
number of iterations. Note that the locking procedure introduces an arbitrariness to the final energy, when
a very small M is used, since the final digits of energy depend on where the locking begins. It is therefore
not recommended to start locking too early in iterations which could trap the orbital solution in a local
mimimum. Finally the quality of resulting orbitals can be checked by carrying out a DMRG calculation
with sufficient renormalized states. Using the default value of switch rst DMRG-SCF usually results in the
orbitals that are good enough to reproduce the CASSCF energy.
9.13.4.4 Complete set of BLOCK options
%casscf
...
dmrg
startM 25 # CASSCF mode: number of re-normalized states for a singlee root

250 # CASCI mode: number of re-normalized states for a single root
maxM 25 # CASSCF mode: number of re-normalized states for a singlee root
250 # CASCI mode: number of re-normalized states for a single root
DryRun false # just create an input for Block
SweepTol 1e-9 # energy tolerance for the sweeps
auto_ordering NOREORDER # auto_ordering is an int. If set to 0
# or the alias NOREORDER, the reordering is skipped.
FIEDLER # (default) let Block optimize the active orbital ordering
GAOPT # let Block optimize the active orbital ordering
# using genetic algorithm
hf_occ 0 # user-defined initial Hartree-Fock occupancy manually
1 # default: initial Hartree-Fock occupancy based on the values of
the one-electron integrals
2 # initial Hartree-Fock occupancy based on the energy ordering
of canonical orbitals
TwoDot_to_OneDot 1 # Switch from two-dot expressions to one-dot

OneDot 0 # Only one-dot expressions. %In CASCI mode only.
TwoDot 0 # Only two-dot expressions. %In CASCI mode only.
switch_rst 1e-3 # Specify the threshold of orbital gradient below which DMRG
swithches to one-dot expression by reading previous wavefunction.
warmup 1 # wilson warm-up type
2, 3 or 4 # n=3 is the default option.
The full configuration space of the n sites next to the system
constitutes the environment states in the warm-up.
The remaining sites use the Hartree-Fock guess occupation
nonspinadapted 0 # default: spin-adapted DMRG
1 # non-spin-adatped DMRG in which the spin-density calculation
is available
# Define a schedule for DMRG

MaxIter 14 # Specify maximum number of iterations
NSchedule -1 # default sweep schedule in CASSCF mode

0 # default sweep schedule in CASCI mode
>0 # Number of manual sweep schedule parameters
# All schedule parameters must be set if this flag is set manually!
sche_iteration 0, 4, 8 # vector with sweep-number to execute changes
# (schedule parameter)
sche_M 50,100,500 # vector with corresponding M values (schedule parameter)

sche_sweeptol 1e-4,1e-6,1e-9 # vector with sweep tolerances (schedule parameter)
sche_noise 1e-8, 1e-11,0.0 # vector with the noise level (schedule parameter)
# Define a separate maxM for DMRG-NEVPT2

nevpt2_maxm 25 # set maximum number of renormalized states
for DMRG-NEVPT2 calculation (default: MaxM)
end
end
9.13.5 Appendix: Porphine -active space calculation
We provide a step-by-step basis on localizing the -orbitals of the porphine molecules and running a CASSCF-
DMRG calculation on this system. It will be important to obtain an initial set of orbitals, rotate the orbitals
which are going to be localized, localize them, and finally run the CASSCF calculation. We will abbreviate
the coordinates as [. . . ] after showing the coordinates in the first input file, but please note they always need
to be included.
1. First obtain RHF orbitals:
# To obtain RHF orbitals

!cc-pvdz
* xyz 0 1
N 2.10524 -0.00000 0.00000
N -0.00114 1.95475 -0.00000
N -2.14882 0.00000 -0.00000
N -0.00114 -1.95475 0.00000
C 2.85587 -1.13749 -0.00000
C 2.85587 1.13749 0.00000
C 1.02499 2.75869 -0.00000
C -1.10180 2.78036 0.00000
C -2.93934 1.13019 -0.00000
C -2.93934 -1.13019 0.00000
C -1.10180 -2.78036 -0.00000
C 1.02499 -2.75869 0.00000
C 4.23561 -0.67410 -0.00000
C 4.23561 0.67410 0.00000

C 0.69482 4.18829 -0.00000
C -0.63686 4.14584 -0.00000
C -4.25427 0.70589 -0.00000
C -4.25427 -0.70589 0.00000
C -0.63686 -4.14584 0.00000
C 0.69482 -4.18829 0.00000
H 5.10469 -1.31153 0.00000
H 5.10469 1.31153 -0.00000
H 1.36066 5.02946 0.00000
H -1.28917 5.00543 0.00000
H -5.12454 1.34852 0.00000
H -5.12454 -1.34852 -0.00000
H -1.28917 -5.00543 -0.00000
H 1.36066 -5.02946 -0.00000
C 2.46219 2.41307 0.00000
C -2.39783 2.44193 0.00000
C -2.39783 -2.44193 -0.00000
C 2.46219 -2.41307 -0.00000
H 3.18114 3.22163 -0.00000
H -3.13041 3.24594 -0.00000
H -3.13041 -3.24594 0.00000
H 3.18114 -3.22163 0.00000
H 1.08819 0.00000 -0.00000
H -1.13385 -0.00000 0.00000
*
2. We then swap orbitals with -character so they are adjacent to each other in the active space. (
orbitals are identified by looking at the MO coefficients). When they are adjacent in the active space,
they can be easily localized in the next step.
#To rotate the orbitals (so that we can localize them in the next step)
!cc-pvdz moread noiter
%moinp "porphine.gbw"
%scf
rotate
{70, 72, 90}
{65, 71, 90}
{61, 70, 90}
{59, 69, 90}
{56, 68, 90}
{88, 84, 90}
{92, 85, 90}
{93, 86, 90}
{96, 87, 90}
{99, 88, 90}

{102, 89, 90}
{103, 90, 90}
{104, 91, 90}
end
end
* xyz 0 1
[...]
*
3. After rotating the orbitals, we localize the 13 occupied -orbitals. This is performed using the orca loc
code. The input file follows.
porphine_rot.gbw
porphine_loc.gbw
0
68
80
120
1e-3
0.9
0.9
1
4. After localizing the occuppied orbitals, we localize the 11 virtual -orbitals using the orca loc code
once again. The input file follows.
porphine_loc.gbw
porphine_loc2.gbw
0
81
91
120
1e-3
0.9
0.9
1
5. After these steps are complete, we run a CASSCF-DMRG calculation. The standard input file is shown
below
!cc-pvdz moread pal4

%moinp "porphine_loc2.gbw"
%MaxCore 16000
%casscf nel 26
norb 24
nroots 1
CIStep DMRGCI
end
* xyz 0 1
[...]
*
9.14 Relativistic Options
The relativistic methods in ORCA are implemented in a fairly straightforward way but do require some
caution from the user. The options are controlled through the %rel block which features the following
variables:
%rel
#----------------------------------------------------
# Basic scalar relativistic method
#----------------------------------------------------
method DKH # Douglas-Kroll-Hess
ZORA # ZORA (numerical integration)
IORA # IORA (numerical integration)
IORAmm # IORA with van Wuellens
# modified metric
ZORA_RI # ZORA (RI approximation)
IORA_RI # IORA (RI approximation)
IORAmm_RI # IORA (RI approximation)
# and modified metric
# ---------------------------------------------------
# Choice of the model potential for ALL methods
# ---------------------------------------------------
ModelPot VeN, VC, VXa, VLDA
# Flags for terms in the model potential (see eq. 9.198)
# these settings do not have any effect for DKH
# =0 not included =1 included
# WARNING: default is currently 1,0,0,0 for ZORA and IORA and
# VeN = nuclear attraction term
# VC = model Coulomb potential
# VXa = model Xalpha potential
# VLDA= VWN-5 local correlation model pot.
Xalpha 0.7 # default value for the X-Alpha potential,
# only has an effect when VXa is part of the model potential
# --------------------------------------------------
# This variable determines the type of fitted atomic
# density that enters the Coulomb potential part of the
9.14 Relativistic Options 449
# model potential (has no effect when using DKH):

# --------------------------------------------------
ModelDens rhoDKH # DKH4 model densities (default)
rhoZORA # ZORA model densities
rhoHF # Hartree-Fock model densities
# --------------------------------------------------
# This flag controls whether only one center terms
# retained. If this is true an approximate treat-
# ment of relativistic effects result but geom-
# try optimizations CAN BE PERFORMED WITH ALL
# METHODS AND MODEL POTENTIAL
# In addition one gets NO gauge noninvariance
# errors in ZORA or IORA
# --------------------------------------------------
OneCenter false # default value
# --------------------------------------------------
# Specify the speed of light used in relativistic
# calculations
# --------------------------------------------------
C 137.0359895 # speed of light used (137.0359895 is the default value)
# synonyms for C are VELIT, VELOCITY
# --------------------------------------------------
# Picture change for properties
# ---------------------------------------------------
PictureChange 0 # (or false): no picturechange (default)
1 # (or true): include picturechange
2 # for DKH: use second-order DKH transformation of the
# SOC operator (see section 9.23.2.7)
# ---------------------------------------------------
# Order of DKH treatment (this has no effect on ZORA calculations)
# ---------------------------------------------------
order 1 # first-order DKH Hamiltonian
2 # second-order DKH Hamiltonian
# ---------------------------------------------------
# Kind of Foldy-Wouthuysen transformation for picturechange effects
# in g tensors (see section 9.14.4)
# ---------------------------------------------------
fpFWtrafo true # do not include vector potential into momentum (default)
false # include vector potential
# ---------------------------------------------------
# Finite Nucleus Model: (see section 9.14.5)
# ---------------------------------------------------
FiniteNuc false # Use point-charge nuclei (default)
true # Use finite nucleus model
end
9.14.1 Approximate Relativistic Hamiltonians
In the relativistic domain, calculations are based on the one-electron, stationary Dirac equation in atomic
units (rest mass subtracted)
hD = ( 1) c2 + c p + V = E.

(9.182)
The spinor can be decomposed in its so-called large and small components
!
L
= (9.183)
S
These are obviously coupled through the Dirac equation. More precisely, upon solving for S , the following
relation is obtained:
1
1 EV
S = 1+ pL = RL (9.184)
2c 2c2
Through the unitary transformation
!
+ R+ 1
U= with + = , = 1 ,
R+ 1+R+ R 1+RR+
the Hamiltonian can be brought into block-diagonal form
!
++
h 0
U + hD U = (9.185)
0
h
The (electronic) large component thus has to satisfy the following relation
h++ L = + h++ + h R + R+ (h + h R) + L = E+ L .

(9.186)
The approximate relativistic schemes implemented in ORCA use different methods to substitute the exact
relation 9.186 with approximate ones.
Two approximation schemes are available in ORCA: the regular approximation and the Douglas-Kroll-Hess
(DKH) approach.
9.14.2 The Regular Approximation
In the regular approximation, 9.186 is approximated by
c
R= p. (9.187)
2c2 V
At the zeroth-order level (ZORA), = 1, so that the ZORA transformation is simply
!
1 R+
UZORA = (9.188)
R 1
and the corresponding Hamiltonian given by
ZORA = V + c p 1
h ++ c p. (9.189)
2c2 V
At the infinite-order level (IORA), is taken into account, so that
!
+ 0
UIORA = UZORA (9.190)
0
and

IORA 1
h++ = + V + c p c p + (9.191)
2c2 V
is the corresponding Hamiltonian. Note that despite the name infinite-order regular approximation this
is still not exact.
In ORCA, the spin-free (scalar-relativistic) variant of ZORA and IORA are implemented. These are obtained
from those above through the replacement
1 1
p pp 2 p. (9.192)
2c2 V 2c V
The regular Hamiltonians contain only part of the Darwin term and no mass-velocity term. A problem with
relations 9.191 and 9.189 is that due to the non-linear dependence of the resulting regular Hamiltonians on V ,
a constant change of V , which in the Dirac and Schrodinger equations will result in a corresponding change
of energy
E E + const (9.193)
does not so in the regular approximation. Several attempts have been made to circumvent this problem.
The scaled ZORA variant is one such procedure. Another one is given through the introduction of model
potentials replacing V . Both approaches are available in ORCA.
The scaled ZORA variant
This variant goes back to van Lenthe et al. [367]. The central observation is that the Hamiltonian
hZORA
hscaledZORA = D E (9.194)
1
1 + L c p (2c2 V 2 c p L

)
produces constant energy-shifts E E +const when the potential V is changed by a constant for hydrogenic
ions. For many-electron systems, the scaled-ZORA Hamiltonian still does not yield simple, constant energy
shift for V V + const. But it produces the exact Dirac energy for hydrogen-like atoms and performs better
than the first-order regular approximation for atomic ionization energies.
The regular approximation with model potential
The idea of this approach goes back to Van W

ullen [123], who suggested the procedure for DFT. However
we also use it for other methods. The scalar relativistic ZORA self-consistent field equation is in our
implementation (in atomic units):
c2

p p + Veff i = i i (9.195)
2c2 V
where c is the speed of light. It looks like the normal nonrelativistic KohnSham equation with the KS
potential Veff :
(r0 )
Z
X ZA
Veff (r) = + dr0 + Vxc [] (r) (9.196)
|r RA | |r r0 |
A
(ZA is the charge of nucleus A and RA is its position; (r) is the total electron density and Vxc [] the
exchange-correlation potential the functional derivative of the exchange-correlation energy with respect to
the density). The kinetic energy operator T = 12 2 of the nonrelativistic treatment is simply replaced by
the ZORA kinetic energy operator:
c2
T ZORA = p p (9.197)
2c2 V
Clearly, in the regions where the potential V is small compared to c2 , this operator reduces to the nonrelativistic
kinetic energy. V could be the actual KS potential. However, this would require to solve the ZORA equations
in a special way which demands recalculation of the kinetic energy in every SCF cycle. This becomes expensive
and is also undesirable since the ZORA method is not gauge invariant and one obtains fairly large errors
from such a procedure unless special precaution is taken. Van W ullen [123] has therefore argued that it is a
reasonable approximation to replace the potential V with a model potential Vmodel which is constructed as
follows:
model (r0 ) 0
Z
X ZA
Vmodel = LDA
model
+ dr + Vxc (r) (9.198)
|r RA | |r r0 |
A
The model density is constructed as a sum over spherically symmetric (neutral) atomic densities:
X
model (r) = A (r) (9.199)
A
Thus, this density neither has the correct number of electrons (for charged species) nor any spin polarization.
Yet, in the regions close to the nucleus, where the relativistic effects matter, it is a reasonable approximation.
The atomic density is expanded in a sum of s-type gaussian functions like:
X
2
A (r) = di exp i |r RA | (9.200)
i
The fit coefficients were determined in three different ways by near basis set limit scalar relativistic atomic
HF calculations and are stored as a library in the program. Through the variable ModelDens (vide supra)
the user can choose between these fits and study the dependence of the results in this choice (it should be
fairly small except, perhaps, with the heavier elements and the HF densities which are not recommended).
The individual components of the model potential (eq. 9.198) can be turned on or off through the use of the
variable ModelPot (vide supra).
Van W ullen has also shown that the calculation of analytical gradients with this approximation becomes close
to trivial and therefore scalar relativistic all electron geometry optimizations become easily feasible within
the ZORA approach. However, since T ZORA is constructed by numerical integration it is very important that
the user takes appropriate precaution in the use of a suitable integration grid and also the use of appropriate
basis sets! In the case of OneCenter true the numerical integration is done accurately along the radial
coordinate and analytically along the angular variables such that too large grids are not necessary unless
your basis set is highly decontracted and contains very steep functions.
9.14.3 The Douglas-Kroll-Hess Method
The Douglas-Kroll-Hess (DKH) method expands the exact relation 9.186 in the external potential V. In
ORCA the first- and second-order DKH methods are implemtented. The first-order DKH Hamiltonian is
given by
(1) = Ep + Ap V Ap + Bp V (p) Bp ,
h (9.201)
++
with
s
p E p + c2 c
EP = c4 + c2 p2 , Ap = , Bp = p (9.202)
2Ep 2Ep (Ep + c2 )
At second order, it reads
h (1) + 1 [Wp , O]
(2) = h (9.203)
++ ++
2
where
cp
{Wp , Ep } = O, O = Ap [Rp , V ] Ap , Rp = (9.204)
E p + c2
(2) is implemented.
define the second-order contribution. In ORCA, the spin-free part of h ++
The occurrence of the relativistic kinetic energy, EP , which is not well-defined in position space, makes a
transformation to the p2 -eigenspace necessary. Thus any DKH calculation will start with a decontraction of the
basis set, to ensure a good resolution of the identity. Then the non-relativistic kinetic energy is diagonalized
and the EP -dependent operators calculated in that space. The potential V and V (p) are transformed to
p2 -eigenspace. After all contributions are multiplied to yield the (first- or second-order) Hamiltonian, the
transformation back to AO space is carried out and the basis is recontracted.
The (spin-free) DKH-Hamiltonians contain all spin-free, relativistic correction terms, e.g. the mass-velocity
and Darwin terms. As the potential enters linearly, no scaling or model potential is necessary to introduce
the correct behaviour of the energy under a change
V V + const. (9.205)
In all these respects the DKH Hamiltonians are much cleaner than the regular Hamiltonians.
9.14.4 Picture-Change Effects
Irrespective of which Hamiltonian has been used in the determination of the wave function, the calculation of
properties requires some special care. This can be understood in two ways: First of all, we changed from
the ordinary Schrodinger Hamiltonian to a more complicated Hamiltonian. As properties are defined as
derivatives of the energy, it is clear that a new Hamiltonian will yield a new expression for the energy and
thus a new and different expression for the property in question. Another way of seeing this is that through
the transformation U , we changed not only the Hamiltonian but also the wave function. To obtain the
property at hand as the expectation value of the property operator with the wave function, we have to make
sure that property operator and wave function are actually given in the same space. This is done through a
transformation of either the property operator or the wave function.
In any case, the difference between the non-relativistic and (quasi) relativistic property operator evaluated
between the (quasi) relativistic wave function is called the picture-change effect. From what was said above,
this is clearly not a physical effect. It describes how consistent the quasi relativistic calculation is carried out.
A fully consistent calculation requires the determination of the wave function on the (quasi) relativistic level
as well as the use of the (quasi) relativistic property operator. This is obtained through the choice
%rel picturechange true end
in the %rel block. It may be that the (quasi) relativistic and non-relativistic property operator do produce
similar results. In this case, a calculation with picture changes turned off (PictureChange false) may
be a good approximation. This is, however, not the rule and cannot be predicted before carrying out the
calculation. It is therefore highly recommended to set PictureChange true in all (quasi) relativistic property
calculations! Consistent picture-change effects on the DKH2 level have been implemented for the g-tensor
and the hyperfine coupling tensor. Using
%rel picturechange 1 end
only first-order changes on the property operators are taken into account. This reduces the computational
cost of course. But since this is in no way a significant reduction, this choice is not recommended.
For magnetic properties, the DKH transformation and consequently the DKH Hamiltonian and the corre-
sponding property operators are not unique. Depending on whether the magnetic field is included in the
free-particle FoldyWouthuysen (fpFW) transformation carried out in the first step of the DKH protocol
or not, two different Hamiltonians result. If the magnetic field is included in the fpFW transformation, the
resulting Hamiltonian is a function of the gauge invariant momentum
= p + A. (9.206)
It is therefore gauge invariant under gauge transformations of the magnetic vector potential A and thus are
the property operators derived from it. This is referred to as fFW DKH Hamiltonian. If the magnetic field
is not included in the FW transformation, the resulting Hamiltonian is a function of the kinetic momentum
p only and thus is not gauge invariant. The latter Hamiltonian is referred to as fpFW DKH Hamiltonian. A
comparison of both Hamiltonians is given in Table 9.12.
Table 9.12: Comparison of the properties of the fpFW and fFW

DKH Hamiltonians. For details see Ref. [368].
Criterion fpFW Hamiltonian fFW Hamiltonian
Convergence of Eigenvalues
? yes
to Dirac Eigenvalues
1st order is bounded no yes
Reproduces Pauli Hamiltonian no yes
Gauge invariance no yes
Lorentz invariance no no
From this Table, it becomes clear that the fFW DKH Hamiltonian is clearly preferred over the fpFW
Hamiltonian. To obtain the property operators, it is however necessary to take the derivatives of these
Hamiltonians. It turns out that in the case of the hyperfine-coupling tensor, the necessary derivatives produce
divergent property operators in the case of the fFW DKH Hamiltonian. This may be due to the unphysical
assumption of a point-dipole as a source of the magnetic field of the nucleus. As a physical description of the
magnetization distribution of the nucleus is not available due to a lack of experimental data, the magnetization
distribution is assumed to be the same as the charge distribution of the nucleus, see Section 9.14.5. This is
unphysical as the magnetization is caused by the one unpaired nucleon in the nucleus whereas the charge
distribution is generated by the protons in the nucleus. So, physically, the magnetization should occupy
a larger volume in space than the charge. This might also be the reason why the resulting finite-nucleus
model is insufficient to remedy the divergencies in the fFW hyperfine-coupling tensor. Consequently, the
hyperfine-coupling tensor is only implemented in the version resulting from the fpFW DKH Hamiltonian. In
the case of the g-tensor both versions are implemented and accessible via the keyword
%rel fpFWtrafo true/false end
By default, this keyword is set to true. A detailed form of the property operators used for the g-tensor and
hyperfine-tensors can be found in Ref. [368].
9.14.5 Finite Nucleus Model
Composite particles like nuclei have, as opposed to elementary particles, a certain spatial extent. While the
point-charge approximation for nuclei is in general very good in nonrelativistic calculations, in relativistic
calculations it might lead to nonnegligible errors. A finite-nucleus model is available for all calculations in
the ORCA program package. It is accessible from the %rel block via
%rel FiniteNuc true/false end
By default, this keyword is set to false. If the keyword is set to true, finite-nucleus effects are considered in
the following integrals:
nucleus potential V
DKH-integral V (p)
one-electron spin-orbit integrals SOC (also in one-electron part of SOMF)
electric-field gradient EFG (and thus, as a consequence in the Fermi-contact and spin-dipole terms of
the hfc tensor)
nucleus-orbit integral NUC
angular-momentum integral l
The finite-nucleus model implemented in ORCA is the Gaussian nucleus model of Ref. [369].
9.14.6 Basis Sets in Relativistic Calculations
For relativistic calculations, special basis sets have been designed, both as DKH and ZORA recontractions of
the non-relativistic Ahlrichs basis sets (in their all-electron versions) for elements up to Xe, and as purpose-
built segmented all-electron relativistically contracted (SARC) basis sets for elements beyond Xe [59]. Their
names are ZORA- or DKH- followed by the conventional basis set name. See section 9.3 for a complete
list.
NOTES:
9.15 Approximate Full CI Calculations in Subspace: ICE-CI 457
It is important to recognize that in the one-center approximation (OneCenter true) ALL methods
can be used for geometry optimization. Several papers in the literature show that this approximation
is fairly accurate for the calculation of structural parameters and vibrational frequencies. Since
this approximation is associated with negligible computational effort relative to the nonrelativistic
calculation it is a recommended procedure.
The ZORA/RI, IORA/RI and IORAmm/RI methods are also done with the model potential. Here
we do the integrals analytically except for the XC terms which has clear advantages. However, the
RI approximation is performed in the actual orbital basis sets which means that this set has to be
large and flexible. Otherwise significant errors may arise. If the basis sets are large (ZORA/RI) and
the numerical integration is accurate (ZORA), the ZORA and ZORA/RI (or IORA and IORA/RI)
methods must give identical within to Eh accuracy.
9.15 Approximate Full CI Calculations in Subspace: ICE-CI
9.15.1 Introduction
In many circumstances, one would like to generate a wavefunction that is as close as possible to the full-CI
result, but Full CI itself is out of the question for computational reasons. Situations in which that may be
desirable include a) one wants to generate highly accurate energies for small molecules or b) one wants to
sort out a number of low-lying states or c) one wants to run CASSCF calculations with larger active spaces
than the about fourteen orbitals that have been the state of the art for a long time.
ORCA features a method that has been termed Iterative-Configuration Expansion Configuration Interaction
(ICE-CI). It is based on much older ideas brought forward by Jean-Paul Malrieu and his co-workers in the
framework of the CIPSI (an abbreviation for a method with a rather bulky name Configuration Interaction
by Perturbation with multiconfigurational zeroth-order wave functions Selected by Iterative process) in the
early 1970s.
The goal of the ICE-CI is to provide compact wavefunction(s) (e.g. one or several states) close to the full-CI
limit at a small fraction of the computational cost. However, ICE-CI itself is not designed to deal with
hundreds of atoms or thousands of basis functions. Thus, unlike, say DLPNO-CCSD(T) which is a high
accuracy method for treating large sytems, ICE-CI is either a highly robust high accuracy method for very
small systems or a building block for large systems. By itself it can treat a few dozen electrons and
orbitals e.g. much more than full CI but it cannot do wonders. Its scope is similar to the density matrix
renormalization group (DMRG) or Quantum Conte Carlo Full CI (QMCFCI) procedures.
ICE-CI should be viewed as a multireference approach. It is self-adaptive and robust, even in the presence
of near or perfect degeneracies. It yields orthogonal states (when applied to several states) and spin
eigenfunctions. It also yields a density and a spin density.
9.15.2 The ICE-CI and CIPSI Algorithms
The general idea of ICE-CI is straightforward: Consider a many-particle state that has an at least sizeable
contribution from a given configuration n0 (this is a set of occupation numbers for the active orbitals that are
n0p = 0, 1 or 2 (p = any active orbital). By nature of the non-relativisitic Hamiltonian only configurations
that differ by at most two orbital occupations from n0 . We can use perturbation theory to select the subset
of singles- and doubles that interact most strongly with n0 and then solve the variational problem. We
can then analyze the CI vector for configurations that make a dominant contribution to the found state.
Say, we single out the configurations with CI2 > Tgen . This defines the generator set of configurations.
The other configurations are called variational configurations. They are treated to infinite order by the
variational principle, but are not important enough to bring in their single- and double excitations. In the
next iteration, we perform singles- and doubles relative to these general configurations and select according
to their interaction with the dominant part of the previous CI vector (truncated to the generators). This
procedure can be repeated until no new important configurations are found and the total energy converges
Flowchart
(See Figure 9.6).
Solve HC=EC in nSD
user defined or
automatically generated Find generator 2
start configuration(s) n configurations
CI >Tgen
Single+Double Single+Double
Just two thresholds
Excitations n+nSD excitations on generators
Tgen & Tvar
nV+nSD
Select on nSDnV 2
individually Select on nSDnV

contracted
1
I
I S +D | H | C
I gen
I
I >Tvar
2
1
I
I S +D | H | 0 >Tvar
Check energy or
configuration convergence
Final solve
no yes HC=EC
=optional Densities,
Figure 9.6: Flowchart of the ICE-CI procedure.
The described procedure is very similar to Malrieus three level CIPSI procedure. One major technical
difference, is that ICE is centered around configurations rather than determinants. A configuration is a
set of occupation numbers 0, 1 or 2 that describes how the electrons are distributed among the available
spatial orbitals. A configuration state function (CSF) is created by coupling the unpaired spins in a given
configuration to a given total spin S. In general there are several, if not many ways to construct a linearly
independent set of CSFs. CSFs on the other hand can be expanded in terms of Slater determinants, but
there are more Slater determinants to a given configuration than CSFs. For example for a CAS(14,18)
calculation one has about 109 determinants, but only about 3x108 CSFs and 3x107 configurations. However,
all logic happens at the level of configurations. That is, it is the relationship between two configurations
that determines whether and if yes, by which integrals the CSFs or determinants of two given configurations
interact. Since the configuration space is so much more compact than the determinant space substantial
computational benefit can be realized by organizing the calculation around the concept of a configuration. In
general, in ICE-CI all CSFs that belong to a given configuration are included and all selection quantities are
summed over all CSFs of a given configuration before it is decided whether this CSFs is included or not.
It should be noted that although the procedure contains a perturbative element, the final energy is strongly
dominated by the variational energy and hence, for all intents and purposes, the ICE-CI procedure is
variational (but not rigorously size consistent size consistency errors are on the same order of magnitude as
the error in absolute energy).
9.15.3 A Simple Example Calculations
Let us look at a simple calculation on the water molecule:
#
# Check the ICECI implementation
#
! SV
%ice nel 10 # number of active electrons

nroots 1 # number of requested roots
integrals exact # exact 4-index transformation
# can be set to RI to avoid bottlenecks
Tgen 1e-04 # value for Tgen. Default is 1e-4

Tvar 1e-11 # value for Tvar. Default is 1e-11 (1e-7*Tgen)
etol 1e-06 # energy convergence tolerance

end
* int 0 1
O 0 0 0 0.0 0.000 0.000
H 1 0 0 1.0 0.000 0.000
H 1 2 0 1.0 104.060 0.000
*
Let us look at the output:
------------------------------------------------------------------------------
ORCA Iterative Configuration Expansion
- a configuration driven CIPSI type approach -
------------------------------------------------------------------------------
(some startup information)
Integral transformations ... done ( 0.3 sec)

Making an initial Aufbau configuration ... done
Performing S+D excitations from 1 configs ... done ( 0.0 sec) NCFG=581
Performing perturbative selection ... done ( 0.0 sec)
# of configurations before selection ... 581
# of configurations after selection ... 187
rest energy (probably not very physical) ... -3.444299e-10
******************************
* CIPSI MACROITERATION 1 *
******************************
# of active configurations = 187
Now calling CI solver (265 CSFs)
(...)
CI SOLUTION:
STATE 0 MULT= 1: E= -76.0458673135 Eh W= 1.0000 DE= 0.000 eV 0.0 cm**-1
0.95841 : 2222200000000
0.00111 : 2122100001001
...
Selecting new configurations ...done ( 0.0 sec)

# of selected configurations ... 187
# of generator configurations ... 67
Performing single and double excitations relative to generators ... done ( 0.0 sec)
# of configurations after S+D ... 13174
Selecting from the generated configurations ... done ( 0.1 sec)
# of configurations after Selection ... 3828
Root 0: -76.045867314 -0.000000061 -76.045867375
(etc.)
******************************
* CIPSI MACROITERATION 4 *
******************************
# of active configurations = 3870
Initializing the CI ...done ( 0.0 sec)

Building coupling coefficients ...done ( 0 sec)
Now calling CI solver (9611 CSFs)
CI SOLUTION:
STATE 0 MULT= 1: E= -76.0539541874 Eh W= 1.0000 DE= 0.000 eV 0.0 cm**-1
0.95101 : 2222200000000
0.00327 : 2221100001100
...
********* CIPSI IS CONVERGED *********

(one final CI)
********************************************
** ICECI Problem solved in 7.5 sec **
********************************************
FINAL CIPSI ENERGIES
Final CIPSI Energy Root 0: -76.053954243 EH
From the output the individual steps in the calculation are readily appreciated. The program keeps cycling
between variational solution of the CI problem, generation of new configurations and perturbative selection
until convergence of the energy is achieved. Normally, this occurs rapidly and rarely requires more than five
iterations. The result will be close to the Full CI result.
Let us look at a H2 O/cc-pVDZ calculation in a bit more detail (See Figure 9.7). The calculation starts out
with a single Hartree-Fock configuration. The first iteration of ICE-CI creates the singles and doubles and
altogether 544 configurations are selected. These singles and doubles bring in about half of the correlation
energy. Already the second iteration, which leads to 73000 selected CSFs provides a result close to the full
CI. At this point up to quadruple excitations from the Hartree-Fock reference have been included. It is well
known that such quadruple excitations are important for the correct behavior of the CI procedure (near size
consistency will come from the part of the quadruple excitations that are products of doubles). However,
only a very small fraction of quadruples will be necessary for achieving the desired accuracy. In the first
iteration the procedure is already converged and provides 99.8% of the correlation energy, using 0.5% of the
CSFs in the full CI space and at less than 0.2% the calculation time required for solving the full CI problem.
Hence, it is clear that near exact results can be obtained while realizing spectacular savings.
Figure 9.7: An ICE-CI calculation on the water molecule in the cc-pVDZ basis (1s frozen)
9.15.4 Accuracy
The accuracy of the procedure is controlled by two parameters Tgen and Tvar Since we have found that
Tvar = 107 Tgen always provides converged results, this choice is the default. However, Tvar can be
set manually. It can be reduced considerably in order to speed up the calculations at the expense of some
accuracy. Our default values are Tgen = 104 and Tvar = 1011 . This provides results within about 1
mEh of the full CI results (roughly speaking, a bit better than CCSDT for genuine closed-shell systems).
During the development of ICE-CI systematic test calculations have been performed using a reference set of
21 full CI energies on small molecules. The convergence pattern of the mean absolute error is shown in Figure
9.8. It is evident from the figure that the convergence of ICE-CI towards the FCI result is very smooth and
that high accuracy can be obtained. In fact, the default settings lead to an accuracy of <1 mEh deviation to
the full-CI result. Eh accuracy can be achieved by further tightening. The achieved accuracy relative to
accurate coupled-cluster results shows that the accuracy of even CCSDTQ can be surpassed by ICE-CI. The
achievable accuracy is only limited by the value of Tgen and much less so by the value of Tvar . Hence, it is
advisable to use a value for Tvar that is essentially converged and control the accuracy of the procedure by
Tgen .
Figure 9.8: Convergence of the ICE-CI procedure towards the full CI results for a test set of 21 full
CI energy. Shown is the RMS error relative to the Full CI results. The corresponding
errors for various coupled-cluster variants is shown by broken horizontal lines.
9.15.5 Scaling behavior
ICE-CI will break the factorial scaling of the full CI problem and scale polynomially. The actual order of
the polynomial scaling is system dependent and accuracy dependent. In order to provide some impression,
consider some calculations on linear polyene chains.
Figure 9.9: Polyene chains used for scaling calculations.
The results are displayed in the Figure 9.10. It is evident from Figure 9.10 that ICE-CI breaks the factorial
scaling of the full CI problem. In fact, for a thresholds of Tgen =104 , 103 and 102 the observed scalings are
approximately O(N8 ), O(N7 ) and O(N6 ) respectively. These numbers will obviously be very system dependent
but should serve as a rough guide. The calculations become quickly much more expensive if Tgen is tightened.
A rule of thumb is that each order of magnitude tightening of Tgen increases the computation time by a
factor of 10. The above calculations have been performed on a simple desktop computer and it was already
possible to solve a CAS(30,30) problem in less than one day of elapsed time using the default thresholds.
Large active spaces will require either loosening of the tresholds or large, more powerful machines.
Figure 9.10: Scaling behavior of ICE-CI for linear polyene chains (Full -electron active space) as
a functions of system size for different generator thresholds.
9.15.6 Accuracy of the Wavefunction
The accuracy of the many particle wavefunction is not straightforward to check. A reasonable measure,
however, is how well it converges towards the exact result for one-electron expectation values. Since every
expectation value can be written in terms of natural orbitals of the one-particle density as:
D E D X E X X D E
=
O o(xi ) = Dpq hp |
o|q i = np p |
o|p

o
pq p
where o(xi ) is an arbitrary one-particle operator, Dpq is the density matrix of the ICE-CI wavefunction, p
are the natural orbitals of the ICE-CI wavefunction and np are their occupations numbers. It is reasonable to
take the deviation of the natural orbital occupation numbers as a measure for wavefunction convergence.
For example, we treat the H2 O/cc-pVDZ problem again. From the results in Figure 9.11 it becomes evident
that the ICE-CI wavefunction is fairly accurate. At the default threshold the occupation numbers agree
to within 103 with the full CI reference numbers, which means that expectation values will be of similar
accuracy. Interestingly, the largest errors occur in the region of the HOMO-LUMO gap, where apparently
all approximate wavefunction approaches tend to depopulate the high lying orbitals too much and put too
much electron density in the low lying empty orbitals. From comparison, it is seen, that the CCSD natural
occupation numbers for this problem are significantly less accurate. Hence, this is evidence that the ICE-CI
wavefunction is properly converging to the right result.
Figure 9.11: Convergence of the ICE-CI natural orbital occupation numbers. The upper panel is
showing the Full CI occupation numbers, the lower panel the deviation of the ICE-CI
values from these exact values. For comparison, the CCSD natural orbital occupation
numbers are also provided.
9.15.7 Potential Energy Surfaces
You can use ICE-CI to scan entire potential energy surfaces. In general, the non-parallelity error along a
potential energy surface is very small. Thus, ICE-CI yields consistent quality throughout the surface.
For example, let us look at the potential energy surface of the N2 molecule (Figure 9.12) a common test
case for quantum chemical methods. There are not too many methods that would disscociate the triple bond
of N2 correctly ICE-CI is one of them. The potential energy surface is entirely smooth and also correctly
behaves in the dissociation limit. Near the minimum it is very close to high-level coupled-cluster methods
that, however, all fail badly as the bond is stretched.
Figure 9.12: Potential energy surface of the N2 molecule in the SV basis. For comparison higher
level coupled-cluster results are also shown.
It is interesting to observe the variations of the ICE-CI wavefunction along the dissociation potential energy
surface. As an example, we look at the dissociation curve of H2 O where both O-H bonds are simultaneously
stretched (Figure 9.13). It is seen that the ICE-CI method is extremely parallel to the full CI curve at all
distances. Hence, the description of the bond remains consistent, even when Hartree-Fock becomes a bad
approximation. The agreement is particularly good if MP2 natural orbitals are used in the ICE-CI procedure.
With the default value of Tgen = 104 and MP2 natural orbitals the error is consistently below 0.2 mEh. For
tighter thresholds, the error is below 0.05 mEh. By contrast, the CCSD(T) method shows relatively large
deviations from the full CI results and also behaves very non-parallel as a function of O-H distance.
It is instructive to analyze the ICE-CI wavefunction along the dissociation pathway (Figure 9.14). It becomes
apparent that the wavefunctions stays compact along the entire surface, even in the dissociation limit, where
the weight of the Hartree-Fock wavefunction drops to less than 25%. Even in this drastic limit, the ICE-CI
wavefunction consists of only about 60000 CSFs, which is very similar to the size of the wavefunction at
Figure 9.13: Non-parallelity error of ICE-CI for the H2 O molecule in the SV basis. Shown is the
deviation from the full CI value as a function of O-H distance (both bonds stretched).
For comparison, the CCSD(T) curve is also shown
.
equilibrium geometry. As the wavefunction becomes more multiconfigurational, the number of generator
configurations goes slightly up from the equilibrium value of 77 to a maximum of 118 and finally 112 at
dissociation. It is also interesting to note that along the entire dissociation pathway no configuration with
more than 8 open shells is generated, which means that no more than quadruple excitations are contained in
the ICE-CI wavefucntion. The number of iterations required in the ICE-CI procedure also stays constant
along the surface at 4 iterations, which impressively shows that a dominant configuration is not necessary for
a successful ICE-CI calculation.
9.15.8 Excited States
ICE-CI can be used to obtain some insight into excited states starting from no knowledge at all. Of course,
the best was to start an excited state calculation is to have some idea which configurations are important for
the low-lying states of the system. If this is not the case, an automated procedure is used. The program
will first generate an Aufbau configuration using the orbitals that are provided on input. Starting from
this Aufbau configuration, single excitations at the configuration level are performed an the Hamiltonian is
diagonalized for the required number of roots. These roots are then analyzed for the leading configurations
and the regular ICE-CI procedure is started from those configurations. For example, look at a calculation on
the CN radical. In this case, we know the relevant orbitals and leading configurations for the lowest four
roots (a doublet ground state, a doublet excited state and a doublet excited state) and hence can
provide them in the input file as shown below.
Figure 9.14: Analysis of the ICE-CI wavefunction along the O-H dissociation pathway.
#
! cc-pVDZ VeryTightSCF
%casscf nel 7
norb 4
nroots 4
mult 2
end
%ice nel 9
norb 26
nroots 4
cimode 3
tvar 1e-11
tgen 1e-4
refs { 2 2 2 2 1 }
{ 2 2 2 1 2 }
{ 2 2 1 2 2 }
{ 2 1 2 2 2 }
end
end
* xyz 0 2
C 0 0 0
N 0 0 1.07
*
The result is shown below. The excitation energies are reasonable but not highly accurate due to the
limitations of the basis set (experimentally the doublet state is at 1.32 eV and the doublet state at 3.22
eV). There is a very slight symmetry breaking In the doublet state that arises from the selection procedure.
It should be noted that the state averaged CASSCF excitation energies are 0.25 eV and 3.18 eV.
STATE 0 MULT= 2: E= -92.4544563186 Eh DE= 0.000 eV 0.0 cm**-1

0.46140 : 22212000000000000000000000
0.38091 : 21222000000000000000000000
0.81854 : 22221000000000000000000000
0.82067 : 22122000000000000000000000
0.40430 : 22212000000000000000000000
0.42974 : 21222000000000000000000000
Below, it is described how to do ICE-CI calculations on excited states if the dominant configurations are not
known.
9.15.9 Tips and Tricks
ICE-CI can be used very fruitfully together with, say, MP2 natural orbitals. This usually results in results
that are closer to full CI results and at the same lead to more compact wavefunctions (it may be called
nICE). The use of MP2 natural orbitals is requested by choosing UseMP2nat true inside the %ice block.
Alternatively, improved virtual orbitals can be used (requested by UseIVOs true). A comparison is shown in
Scheme 9.15. It is evident that the calculations based on the MP2 natural orbitals show an error relative to
full CI that is almost a factor of two smaller than the corresponding result with canonical orbitals while at
the same time the wavefunction is more compact by more than 30%. Hence, the use of MP2 natural orbitals
appears to be a very good idea in conjunction with the ICE-CI procedure. This also holds when MP2 itself is
a bad approximation (for example in the dissociation limit of the H2O molecule as shown above). On the
other hand, the IVOs behave very similar to canonical orbitals and hence, seem to offer fewer advantages.
Figure 9.15: Comparison of MP2 natural orbitals and improved virtual orbitals for the ICE-CI
procedure (H2O molecule, cc-pVDZ basis, equilibrium geometry)
If ICE-CI is used in conjunction with MP2 natural orbitals, there also is the possibility of letting the program
automatically choose the active space (this is called auto-ICE). The general idea is simple we base the
active space on the MP2 natural orbitals and their occupation numbers. All orbitals between occupation
number say 1.98 down to 0.02 will be included in the active space. A relevant input is shown below.
! cc-pVDZ aug-cc-pV6Z/C Auto-ICE

%ice nmin 1.99 nmax 0.01 end
%paras R= 1.0 end
* int 0 1
O 0 0 0 0 0 0
H 1 0 0 { R} 0 0
H 1 2 0 { R} 104 0
*
If we scan along the H2 O dissociation surface one can see that despite changing active spaces, the dissociation
curves are smooth and remain fairly parallel to the full CI dissociation curve. Depending on the tightness
of the thresholds the active space may change from a small 6 electrons in 5 orbitals to a larger 8 electrons
in 7 or 8 orbitals upon dissociation. This is the expected behavior as the -antibonding orbital becomes
more stable along the bond stretching coordinate. Hence, these results are encouraging in as far as in many
situations the program will be able to select a sensible active space without extended input from the user.
Figure 9.16: Automatic active space selection along the H2 O dissociation surface. The reference
curve (blue triangles) is the ICE-CI method for the full orbital space with the default
parameters.
Another place, where automatic selection comes in conveniently is in the calculation of excited states. If
there are no user supplied configurations, what happens is that the program will first choose an Aufbau
reference configuration and then perform all single excitations relative to this configuration. The program
will then diagonalize the Hamiltonian over the this set of configurations to create 0th order approximations
for the chosen number of roots of interest and then initiate the ICE-CI procedure starting from the leading
configurations of these states. Here is an example for the benzene molecule:
! RHF def2-SVPD def2-SVP/C Auto-ICE

%cclib "/Users/neese/prog_c/orca/cclib/orcacc"
%ice nroots 5
nmin 1.98
nmax 0.02
integrals ri
end
* int 0 1
C 0 0 0 0.000000 0.000 0.000
C 1 0 0 1.389437 0.000 0.000
C 2 1 0 1.389437 120.000 0.000
C 3 2 1 1.389437 120.000 0.000
C 4 3 2 1.389437 120.000 0.000
C 5 4 3 1.389437 120.000 0.000
H 1 2 3 1.082921 120.000 180.000
H 2 1 3 1.082921 120.000 180.000
H 3 2 1 1.082921 120.000 180.000
H 4 3 2 1.082921 120.000 180.000
H 5 4 3 1.082921 120.000 180.000
H 6 5 4 1.082921 120.000 180.000
*
(The %cclib statement is explained below and is not mandatory here). The Auto-ICE procedure comes
up with as many as 24 electrons in 19 orbitals, which already is a fairly heavy calculation. The procedure
converges in five iterations and provides indeed the correct states: the gorund state, the 1 B2u state at 6.4 eV,
the 1 B1u state at 8.9 eV and a degenerate 1 E1u state at 10.0 eV. These excitation energies are still in error
by about 2 eV relative to experiment, which is mainly due to missing dynamic correlation. However, the
correct states and their sequence has been found.
The ICE-CI can be used to find the ground state if the actual ground state is not known. To this end, one
simply has to turn off the selection steps. This makes the calculations more expensive, but they will converge
to the lowest state. In the example below (again, the H2 O molecule) we start from a random quintuply
excited configuration the ICE-CI still finds the ground state after four iterations:
%ice nel 8
norb 23
nroots 1
tvar 1e-11
tgen 1e-04
etol 1e-06
# selection
SelStart false
SelIter false
# algorithm details
useivos false
integrals exact
cimaxdim 5 #Davidson expansion space = MaxDim * NRoots
cimode 3
# spatial sym (buggy)

irrep 0
# startup (optional)
refs { 2 1 0 1 0 2 1 0 1 }
end
end
However, if one wants to converge to an excited state, one should turn on the selection. In the example below
(once more the water molecule) one can converge to the second excitated singlet state by judicious choice of
the start configuration:
%ice nel 8
norb 23
nroots 1
tvar 1e-11
tgen 1e-04
etol 1e-06
# selection
SelStart true
SelIter true
# algorithm details
useivos false
integrals exact
cimaxdim 5
cimode 3
# spatial sym (buggy)
irrep 0
# startup (optional)
refs { 2 2 1 2 0 1 }
end
end
9.15.10 Large-scale approximate CASSCF: ICE-SCF
ICE-CI can be used as a replacement for the CI step in a CASSCF framework. In this way, much larger
CASSCF calculations than previously possible can be envisioned. In using the ICE-CI in this way, the active
orbitals should be chosen as natural orbitals in order to ensure a proper canonicalization. In general, ICE-CI
results will not be invariant with respect to the choice of orbitals. However, in practice we have not found
this to be problematic. We refer to this as ICE-SCF.
The use is simple: in the %casscf block choose:
%casscf
...
cistep ice
# optional input with refined settings
ci
tgen 1e-4 # controls accuracy (default = 1e-4)
tvar 1e-11 # default = 1e-7 * TGen
maxiter 100 # number of allowed cycles (default = 64)
end
end
The entire remaining input is the one for standard CASSCF calculations. In this way one can do CASSCF
calculations with very large active space in reasonable turnaround times. We have not observed convergence
problems that are worse than in the standard CASSCF procedure. The results in Figure 9.17 show that the
deviations from regular CASSCF energies are very small. The largest deviation observed for C2 H4 is on the
order of 0.2 mEh, which appears acceptable. Note that the CASSCF tutorial also covers larger examples
and excitations energies computed with the ICE-CI as CI solver. As mentioned in the CASSCF section 9.11,
some feature are not supported for ICE-CI e.g. magnetic properties as well NEVPT2 corrections are not yet
available.
Figure 9.17: Deviations of ICE-SCF from CASSCF energies for a selection of molecules (standard
truncation parameters Tgen = 104 and Tvar = 1011 )
Since CASSCF is fully variational, it is possible to optimize geometries with that procedure. It is our
experience so far, that the ICE-SCF geometries are virtually indistinguishable from CASSCF geometries (an
example is shown in Figure 9.18).
9.15.11 The entire input block explained
For completeness, the parameters that can be specified in the input block are summarized below:
%ice
nel 8 # number of active electrons
nroots 1 # number of roots
mult 1 # requested multiplicity
irrep 0 # requested irrep (buggy :-()
Figure 9.18: CASSCF and ICE-SCF optimized geometries for methylene and ozone (cc-pVDZ basis
set, default parameters).
tgen 1e-04 # generator threshold

tvar -1e-7 # negative -> 1e-7*tgen
etol 1e-06 # convergence tolerance
# algorithm details
useMP2nat false # use MP2 natural orbitals
useivos false # use improved virtual orbitals
useQROs false # For UHF: use quasi-restricted MOs?
integrals exact # exact or ri transformation
cimaxdim 10 # max. size of expansion space in Davidson procedure
cimode 3 # default=accelerated CI, other settings not recommended
# startup configurations(optional)
refs { 2 2 2 2 0 }
{ 2 2 2 0 2 }
end
end
9.15.12 A Technical Note: orca cclib
We should finally mention a technical aspect. The CI procedure in ICE-CI is based around the so-called one
particle coupling coefficients
AIJ q

pq = I|Ep |J (9.207)
where AIJ q
pq is a coupling coefficient, I and J are configuration state functions (CSFs) and Ep is the spin-free
excitation operator that promotes an electron from orbital p to orbital q. The values of these coupling
coefficients only depend on the logical relationship between the CSFs I and J but not on the absolute values
of I, J, p, q. In fact, they only depend on the number of unpaired electrons in I and the total spin S that
both CSFs refer to. Hence, prototype coefficients can be pre-tabulated. This is normally done in a CI run at
the beginning of the run. However, in ICE-CI it may have to be repeated several dozen times and for large
numbers of open shells (say 14), the process is time and memory consuming.
In order to ease the computational burden, we have provided a small utility program that tabulates the
coupling coefficients for a given total spin S (rather the multiplicity M = 2S + 1) and maximum number of
open shells. This program is called orca cclib. It is called like:
orca_cclib Mult MaxNOpen
It will produce a series of files orcacc.el.mult.nopen (electron density coupling coefficients) and orcacc.sp.mult.nopen
(spin-density coupling coefficients) in the current directory. These files are binary files. They can be copied to
an arbitrary directory. You instruct the program to read these coefficients (rather than to recalculate them
all the time) by setting the path to this directory:
# My Job
! def2-SVP Auto-ICE
%cclib "/user/me/orca/cclib/orcacc"
The remaining part of the filename will be automatically added by the program. This option can save
humongous amounts of time. The coupling coefficient library needs to be made for the desired multiplicities
only once. The practical limit will be 14-16 open shells. If you are running the calculation on a cluster using
some submit script, you have to ensure that the provided cclib path is accessible from the compute node.
9.16 CI methods using generated code
The AUTOCI module is replacement of the orca mdci for cases, where manual implementation of the method
would be tedious or practically impossible. The module works with all types of reference wave function
available in ORCA, i.e. RHF, ROHF, UHF and CASSCF and offers CI and related methods. At current
stage, only the energies are available and all computational modules do run only in serial mode, i.e. no
parallelization is implemented. All the methods are implemented in canonical orbital basis and storing all
integrals on disk.
9.16.1 Introduction
All the theories are obtained by the means of automated programming within the ORCA-AGE (Automated
Generator Environment for ORCA). The CI module reads in the SCF wavefunction and optimizes the
coefficient of the CI expansion. Conceptually, the module is very similar to orca mdci, therefore the input
and output do have a lot in common.
9.16.2 Input
All parameters applicable to the autoci module are shown below.

9.16 CI methods using generated code 475
%autoci
# Algorithm selection
citype # Type of the CI expansion to be applied (one of following)
CISD # configuration interaction with single and double substitutions
POLYCI # Polynomial CI
CCSD # Polynomial CI of the order 4 for RHF reference
FICMRCI # Fully internally contracted MRCI
FICMRCEPA0 # Fully internally contracted CEPA0
% FICDDCI3 # FIC-MRCI without the IJAB excitation class
% FICDDCI2 # FIC-DDCI3 without the ITAB and IJTA excitation classes
% FICSORCI # FIC-DDCI3 in the basis of Nat. Orbs. of DDCI2
polycidegree 1 # Degree of the polynomial CI expansion (1=CISD, 2QCISD,

4=CCSD)
# converger details
stol 1e-06 # residue convergence tolerance
maxiter 50 # maximum number of iterations
maxdiis 5 # depth of the DIIS memory
diisstartiter 2 # Apply DIIS starting at iteration 1
denomopt # Type of denominator used for update (one of following)
0 # Orbital energies
1 # Dyall Hamiltonian
2 # Full Born-Oppenheimer Hamiltonian
# CAS settings similar to the CASSCF input

nel 6 # number of active electrons (for CAS)
norb 7 # number of active orbitals (for CAS)
mult 1 # requested multiplicity block
nroots 1 # number of roots for mult block
irrep 0 # requested irrep for mult block
nthresh 1e-6 # Threshold for lin. dependencies in the IC-CSFs basis
D3thresh 1e-14 # Density truncation in D3
D4thresh 1e-14 # Density truncation in D4
# Algorithm details
maxcore 2000 # Maximum memory limit in MB
printlevel 3 # Amount of printing
integraltrafotype 0 # Type of integral transformation
0 # Full canonical
1 # Full using RI (RI basis needed)
2 # Up to three-ext (ibac) using RI
3 # Up to three-ext (tbac) using RI
keepints # Keep the transformed integrals on disk
useoldints # Use the transformed integrals found on disk
% # SORCI settings
% sorci1nthresh 1e-6 # nthresh for the DDCI2 step of SORCI
% sorci1maxiter 10 # maxiter for the DDCI2 step of SORCI
% sorci1stol 1e-3 # stol for the DDCI2 step of SORCI
% natpopcut 1e-5 # Cutoff for nat. orb. truncation after DDCI2 step of
% SORCI
# Property calculations
density # type of density requested
NONE # No density calculation

LINEARIZED # Linear part of the density (exact for CISD and POLY-1)
end
N.B. In the case of a UHF reference, only the CISD (polycidegree 1) and CCSD (polycidegree 4) methods
are available. For a ROHF reference, only CISD (polycidegree 1) and quadratic CCSD (polycidegree 2)
calculations can be performed in the current version.
If one wishes to experiment with the module itself and the reference wavefunction stays constant, it is
possible to store the transformed MO integrals on disk (keepints) and reuse them (useoldints). The program
checks only whether the dimension of the integrals on disk match the problem acutally solved, ie. the user is
responsible for valid data.
9.16.3 Fully internally contracted MRCI
Starting point for any multireference approach is a reference wavefunction that consists of multiple deter-
minants or configurations state functions (CSFs). In many instances this is the complete active space SCF
(CASSCF) wavefunction. In the uncontracted MRCI approach, as implemented in the orca mrci module, the
wavefunction is expanded in terms of excited CSFs that are generated by considering excitations with respect
to all reference CSFs. The methodology scales with the number of reference CSFs and hence is restricted to
small reference spaces. Moreover, the configuration driven algorithm used in orca mrci keeps all integrals in
memory, which further limits the overall size of the molecule.
Internal contraction as proposed by Meyer and Siegbahn avoids these bottlenecks. [370, 371] Here, excited
CSFs are generated by applying the excitation operator to the reference wavefunction as whole. The fully
internally contracted MRCI presented here (FIC-MRCI) uses the same internal contraction scheme as the
FIC-NEVPT2 (aka PC-NEVPT2). The entire methodology as well as a comparison with the conventional
uncontracted MRCI is reported in our article. [372] The general input structure is similar to the CASSCF e.g.
the following example input reads an arbitrary set of orbitals and starts the FIC-MRCI calculation.
!def2-tzvp moread allowrhf noiter nofrozencore

%moinp "start.gbw" # could be from CASSCF
%autoci
citype FICMRCI # Fully internally contracted MRCI (singles,doubles)
FICMRCEPA0 # CEPA0 version of FIC-MRCI
nel 2
norb 2
mult 1,3
nroots 3,1
nthresh 1e-6 # removal of linear dependencies in the IC-CSFs

9.17 Geometry Optimization 477
DavidsonOpt 0 # none (default)

1 # Davidson correction
end
Currently, the program is capable of computing total energies and vertical excitation energies. More features
will be available with future releases. Note that the program is not yet parallelized.
9.17 Geometry Optimization
ORCA is able to calculate equilibrium structures (minima and transition states) using the quasi Newton
update procedure with the well known BFGS update [124129], the Powell or the Bofill update. The
optimization can be carried out in either redundant internal (recommended in most cases) or Cartesian
displacement coordinates. As initial Hessian the user can choose between a diagonal initial Hessian, several
model Hessians (Lindh, Almloef, Schlegel), an exact hessian and a partially exact Hessian (both recommended
for transition state optimization) for both coordinate types. In redundant internal coordinates several options
for the type of step to be taken exist. The user can define constraints via two different paths. He can
either define them directly (as bond length, angle, dihedral or Cartesian constraints) or he can define several
fragments and constrain the fragments internally and with respect to other fragments. The ORCA optimizer
can be used as an external optimizer, i.e. without the energy and gradient calculations done by ORCA.
9.17.1 Input Options and General Considerations
The use of the geometry optimization module is relatively straightforward.18
%method RunTyp Opt # use geometry optimization.

#(equivalent is RunTyp=Geom)
end
# or simply "! Opt" in the keyword line
# details of the optimization are controlled here

%geom
MaxIter 50 # max. number of geometry iterations
# (default is 3N (N = number of atoms), at least 50 )
# coordinate type control
coordsys redundant # New redundant internal coords
redundant_old # old set of redundant internal
18
But that doesnt mean that geometry optimization itself is straightforward! Sometimes, even when it is not expected
the convergence can be pretty bad and it may take a better starting structure to come to a stationary point. In
particular floppy structures with many possible rotations around single bonds and soft dihedral angle modes are
tricky. It may sometimes be advantageous to compute a Hessian matrix at a cheap level of theory and then do
the optimization in Cartesian coordinates starting from the calculated Hessian.
# coords (molecules might explode)

redundant_new # New set of redundant internal coords
# (includes non-covalent bonds, i.e. H-bonds)
deloc # Delocalized internals
cartesian # Cartesian coordinates
# transition state (TS) optimization
TS_search EF # Switch on TS search, EF means
# "eigenvector following"
# alternatively use "! OptTS"
TS_Mode {M 0} end # Choose the mode to follow uphill in the
# TS optimization. {M X}: eigenvector of
# the Hessian with X. lowest eigenvalue
# (start counting at zero) (default: X=0)
# Instead of a mode choose an internal coordinate strongly
# involved in the eigenmode followed uphill
TS_Mode {B 0 1} end # bond between atoms 0 and 1 or
TS_Mode {A 2 1 0} end # angle between atoms 2, 1 and 0 or
TS_Mode {D 3 2 1 0} end # dihedral of atoms 3, 2, 1 and 0
# add or remove internal coordinates from the automatically
# generated set of redundant internal coords
modify_internal
{ B 10 0 A } # add a bond between atoms 0 and 10
{ A 8 9 10 R } # remove the angle defined
# by atoms 8, 9 and 10
{ D 7 8 9 10 R } # remove the dihedral angle defined
end # by atoms 7, 8, 9 and 10
# constrain internal coordinates:
Constraints
{ B N1 N2 value C } # the bond between N1 and N2
{ A N1 N2 N1 value C } # the angle defined by N1, N2
# and N3
{ D N1 N2 N3 N4 value C } # the dihedral defined by N1,
# N2, N3 and N4
{ C N1 C } # the cartesian position of N1
{ B N1 * C} # all bonds involving N1
{ B * * C} # all bonds
{ A * N2 * C } # all angles with N2 as central atom
{ A * * * C } # all angles
{ D * N2 N3 * C } # all dihedrals with N2 and N3 as
# central atoms
{ D * * * * C } # all dihedrals
end
# scan an internal coordinate:
Scan B N1 N2 = value1, value2, N end
# perform constrained optimizations with varying N1-N2-
# distance from value1 up to value2 in N steps;

# works as well for angles (use A N1 N2 N3) and for
# dihedrals (use D N1 N2 N3 N4)
Scan B N1 N2 [value1 value2 value3 ... valueN] end
# perform constrained optimizations with N1-N2-distances
# as given in the list;
# works as well for angles (use A N1 N2 N3) and for
# dihedrals (use D N1 N2 N3 N4)
fullScan true # if !ScanTS is requested, fullScan assures
# that the relaxed surface scan is fully
# carried out before the TS optimization is
# started (Default is false)
# fragment optimization:
# 1. all atoms have to belong to a fragment
# 2. you have to connect the fragments
ConnectFragments
{1 2 C} # constrain the internal coordinates
# connecting fragments 1 and 2
{1 2 C N1 N2}# constrain the internal coordinates
# connecting fragments 1 and 2, the
# fragments are connected via atoms
# N1 and N2
{1 3 O} # optimize the internal coordinates
# connecting fragments 1 and 3
{1 3 O N1 N2}# optimize the internal coordinates
# connecting fragments 1 and 3, the
# fragments are connected via atoms
# N1 and N2
end
# 3. you can constrain the fragment internally
ConstrainFragments # constrain all internal coordinates
{ 1 } # containing only atoms of fragment 1
end
# optimize hydrogens
optimizeHydrogens true
# in the context of a normal optimization all internal
# coordinates not involving any hydrogens are constrained
# in the context of a fragment optimization all internal
# coordinates involving hydrogens are optimized (also in a
# constrained fragment)
# freeze the hydrogen positions with respect to the
# heteroatoms
freezeHydrogens true
# invert the defined constraints, i.e. optimize the
# constraints and constrain the remaining coordinates
# this only works for the redundant internal coordinates

# Cartesian coordinates are not affected by invertConstraints
invertConstraints true # step type control
Step qn # quasi-Newton step
rfo # Rational function step (Default for !Opt)
gdiis # gdiis step
prfo # partitioned RFO step (Default for !OptTS)
UseGDIIS false # use GDIIS step (in Cartesian optimization)
# Default is false.
GDIISStart 1.0 # Gradient at which to start GDIIS algorithm
# (in Cartesian optimization)
GDIISMaxE 10 # number of last steps to use in GDIIS
# algorithm
# Step size control
MaxStep 0.3 # maximum step length in internal coordi-
# nates. Default is 0.3 au
Trust -0.3 # Initial trust radius. Default is -0.3 au
# Trust <0 - use fixed trust radius
# of size -trust. I.e. -0.3 means fix
# the trust radius at 0.3
# Trust >0 - use trust radius update. I.e. 0.3
# means start with trust radius 0.3 and update
# the trust radius after each optimization step
# Convergence tolerances. Note that the calculation is
# only converged if all criteria are fullfilled. All
# values given are default values.
TolE 5e-6 # Energy change (a.u.)
TolRMSG 1e-4 # RMS gradient (a.u.)
TolMaxG 3e-3 # Max. element of gradient (a.u.)
TolRMSD 2e-3 # RMS displacement (a.u.)
TolMaxD 4e-3 # Max. displacement (a.u.)
# keyword for frequently used sets of convergence thresholds
Convergence normal # Default
loose
tight
ProjectTR false # project translation and rotation
# default is false. MUST be false for
# redundant internals
end
Keywords for the control of the Hessian (especially important for the TS optimization):
# initial Hessian control

inhess unit # unit matrix
Read # Hessian in a .hess file (e.g. from
# a previous NumFreq run), this command
# comes with the following:
InHessName "filename.hess" # filename of
# Hessian input file
# these only for redundants
Lindh # Lindhs model Hessian
Almloef # Almloefs model Hessian
Schlegel # Schlegels model Hessian
# additional Hessian control for TS optimization
Calc_Hess true # calculate the Hessian numerically at the beginning
Recalc_Hess 5 # calculate the Hessian at the beginning
# and recalculate it after 5,10,.. cycles
Hybrid_Hess [0 1 5 6] end # calculates a Hybrid Hessian
# exact calculation for
# atoms 0, 1, 5 and 6; works also
# with Calc_Hess and Recalc_Hess
NumHess true # requests use of numerical Hessian
# modification of the internal Hessian
Hess_Internal
{A 3 2 1 D 2.0} # define a diagonal Hessian value of
# 2 Eh/Bohr2 for the angle between
# atoms 3 2 1. This can also be done for
# bonds, dihedrals and Cartesian
# coordinates.) The Hessian values of
# multiple coordinates can be modified
reset 5 # reset the modified internal Hessian values
# after 5 cycles
# The following is only recommended
# after a relaxed surface scan
# in this example of the scan coordinate B 1 0;
# "basename.004.xyz" contains the optimized structure
# of the scan step with highest energy
{B 1 0 C}
XYZ1 "scanName.003.xyz" # the xyz-files of the structures
XYZ2 "ScanName.005.xyz" # next to the highest energy point
GBW1 "ScanName.003.gbw" # the gbw-files of the structures
GBW2 "ScanName.005.xyz" # next to the highest energy
# the gbw-files are optional
end
# Hessian update procedure
Update Powell
Bofill # default for TS optimization
BFGS # default for geometry optimization

# Hessian modification (only for P-RFO step)
HESS_Modification Shift_Diag # shift the diagonal elements
# (default)
EV_Reverse # reverse the
# diagonal elements
# Minimal value of Hessian eigenvalues (only P-RFO step)
HESS_MinEV 0.0001 # if an absolute Hessian eigenvalue
# is smaller than this value, it is
# set to HESS_MinEV
# Rebuilding the model Hessian after a number of cycles can
# accelerate the convergene of the optimization
NResetHess 20 # Set the number of geometry steps after which
# a new model Hessian is built (only with BFGS
# update)
NStepsInResetHess 5 # since previous steps and gradients are
# available, it is possible to include
# information about the PES in the
# newly built Hessian (via a BFGS
# update). This number should be
# smaller than NResetHess
end
As for parameter scan runs ORCA has some special options that may help to speed up the optimization:
%geom UseSOSCF false # switches the converger to SOSCF

# after the first point. SOSCF may
# converge better than DIIS if the
# starting orbitals are good.
# default = false
ReducePrint true # reduce printout after the first
# point default=true
# the initial guess can be changed after the first
# point. The default is MORead. The MOs of the pre-
# vious point will in many cases be a very good guess
# for the next point. In some cases however, you may
# want to be more conservative and use a general guess.
OptGuess = OneElec # the one electron matrix
= Hueckel # the extended Hueckel guess
= PAtom; # the PAtom guess
= Pmodel # the PModel guess
= MORead # MOs of the prev. point (default)
end
There are three types of internal coordinates: redundant internals, old redundant internals (redundant old)
and (nonredundant) delocalized internals. All three sets work with the same primitive space of internal
coordinates (stretches, bends, dihedral angles and improper torsions). Only the redundant internals works
with one more type of bends in cases where a normal bend would have been approximately 180 . In redundant
internal coordinates the full primitive set is kept and the Hessian and gradient are transformed into this
potentially large space. A geometry optimization step requires, depending on the method used for the
geometry update, perhaps a diagonalization or inversion of the Hessian of dimension equal to the number of
variables in the optimization. In redundant internal coordinates this space may be 2-4 times larger than the
nonredundant subspace which is of dimension 3Natoms 6(5). Since the diagonalization or inversion scales
cubically the computational overhead over nonredundant spaces may easily reach a factor of 864. Thus,
in redundant internal coordinates there are many unnecessary steps which may take some real time if the
number of primitive internals is greater than 1000 or so (which is not so unusual). The timing problem may
become acute in semiempirical calculations where the energy and gradient evaluations are cheap.
In order to deal with this situation, there is a second method implemented in ORCA which is based on
(nonredundant) delocalized internal coordinates. If everything is well, the two methods show identical
convergence since they start from the same Hessian and finally they span the same space. However, the
delocalized internals take significantly less time for the geometry update since the dimension of the matrices
to be manipulated is only the required 3Natoms 6(5) and there are no expensive transformation. The (only)
drawback of the delocalized internals is that the iterative back-transformation from internal to Cartesian
coordinates is in our experience less robust than in redundant internal coordinates and may fail. In
this case you have to revert to redundant internals. If everything goes well, delocalized internals are a good
choice.
In order to appreciate these differences we briefly outline the theoretical background which is not difficult to
understand:
Suppose, we have a set of nI (redundant) primitive internal coordinates q constructed by some recipe and a
set of nC = 3Natoms Cartesian coordinates x. The B-matrix is defined as:
qi
Bij = (9.208)
xj
This matrix is rectangular. In order to compute the internal gradient one needs to compute the generalized
inverse of B. However, since the set of primitive internals is redundant the matrix is rank-deficient and one
has to be careful. In pratice one first computes the nI nI matrix G:
G = BBT (9.209)
The generalized inverse of G is denoted G and is defined in terms of the eigenvalues and eigenvectors of
G:
!T ! !
U 1 0 U
G = (9.210)
R 0 0 R
Here U are the eigenvectors belonging to the nonzero eigenvalues which span the nonredundant space
and R are the eigenvectors of the redundant subspace of the primitive internal space. If the set of primitive
internals is carefully chosen, then there are exactly 3Natoms 6(5) nonzero eigenvalues of G. Using this
matrix, the gradient in internal coordinates can be readily computed from the (known) Cartesian gradient:
gq = G Bgx (9.211)
The initial Hessian is formed directly in the redundant internal space and then itself or its inverse is updated
during the geometry optimization.
Before generating the Newton step we have to ensure that the displacements take place only in the nonredun-
dant part of the internal coordinate space. For this purpose a projector P 0 :
P0 = GG = G G (9.212)
is applied on both the gradient and the Hessian:
q = P0 gq
g (9.213)
q = P0 Hq P0 + (1 P0 )
H (9.214)
The second term for H sets the matrix elements of the redundant part of the internal coordinate space to
very large values ( = 1000).
Coordinate steps
A Quasi-Newton (QN) step is the simplest choice to update the coordinates and is given by:
1
q = H q gq (9.215)
A more sophisticated step is the rational function optimization step which proceeds by diagonalizing the
augmented Hessian:
! ! !
Hq gq q q
=v (9.216)
gq 0 1 1
The lowest eigenvalue 0 approaches zero as the equilibrium geometry is approached and the nice side effect of
the optimization is a step size control. Towards convergence, the RFO step is approaching the quasi-Newton
step and before it leads to a damped step is taken. In any case, each individual element of q is restricted to
magnitude MaxStep and the total length of the step is restricted to Trust. In the RFO case, this is achieved
by minimizing the predicted energy on the hypersphere of radius Trust which also modifies the direction of
the step while in the quasi-Newton step, the step vector is simply scaled down.
Thus, the new geometry is given by:
qnew = qold + q (9.217)
However, which Cartesian coordinates belong to the new redundant internal set? This is a somewhat
complicated problem since the relation between internals and Cartesians is very nonlinear and the step
in internal coordinates is not infinitesimal. Thus, an iterative procedure is taken to update the Cartesian
coordinates. First of all consider the first (linear) step:
x = Aq (9.218)
with A = BT G . With the new Cartesian coordinates xk+1 = xk + x a trial set of internals qk+1 can be
computed. This new set should ideally coincide with qnew but in fact it usually will not. Thus, one can refine
the Cartesian step by forming
q = qnew qk+1 (9.219)
which should approach zero. This leads to a new set of Cartesians x0 = Aq which in turn leads to a
new set of internals and the procedure is iterated until the Cartesians do not change and the output internals
equal qnew within a given tolerance (107 RMS deviation in both quantities is imposed in ORCA).
Now, consider working with the nonredundant subspace qD = Uq. In this case, the whole procedure becomes
rather simple. The delocalized B-matrix is:
BD = UT B (9.220)
The corresponding G matrix is not rank deficient and can be readily inverted. Then, the gradient in
delocalized internals is calculated as:
gD = G
D BD gc (9.221)
The initial (diagonal) Hessian in q-space is converted to qD simply by UT H0 U. Then the update procedure
and perhaps the diagonalization in the RFO step only works in the nonredundant 3N 6(5) subspace instead
of the full primitive internal space which is an advantage. The back-transformation of the step in delocalized
internals to Cartesians is completely analogous to the redundant internal case. However, in my experience it
is less robust than in the latter case. In principle, the nonredundant subspace changes in every iteration since
the B-matrix changes in every iteration. However, if one would take this into account the whole computational
advantage of the delocalized internals is lost. Consequently, the delocalized internals are kept fixed and unless
the geometry undergoes a qualitative change during the optimization this is not expected to have a large
influence on the convergence characteristics. The final geometries predicted with any type of coordinates
should of course always be the same.
Constrained Optimization
Constraints on the redundant internal coordinates can be imposed by modifying the above projector P 0 with
a projector for the constraints C:
1
P = P0 P0 C (CPC) CP0 (9.222)
C is a diagonal matrix with 1s for the constraints and 0s elsewhere. The gradient and the Hessian are
projected with the modified projector:
gq = P gq (9.223)
q = P Hq P + (1 P )
H (9.224)
Constrained Fragments Optimization
The constrain fragments option was implemented in order to provide a convenient way to handle constraints
for systems consisting of several molecules. The difference to a common optimization lies in the coordinate
setup. In a common coordinate setup the internal coordinates are built up as described in the following:
In a first step, bonds are constructed between atom pairs which fulfill certain (atom type specific) distance
criteria. If there are fragments in the system, which are not connected to each other (this is the case when
there are two or more separate molecules), an additional bond is assigned to the nearest atom pair between
the nonbonded fragments. All other internal coordinates are constructed on the basis of this set of bonds.
Here, in a second step, bond angles are constructed between the atoms of directly neighboured bonds. If
such an angle reaches more than 175 , a special type of linear angles is constructed. In a third step, dihedral
angles (and improper torsions) are constructed between the atoms of directly neighboured angles.
If the constrain fragments option is switched on, the set of bonds is constructed in a different way. The user
defines a number of fragments. For each fragment a full set of bonds (not seeing the atoms of the other
fragments) is constructed as described above. If using this option, the user also has to define which fragments
are to be connected. The connection between these fragments can either be user-defined or automatically
chosen. If the user defines the connecting atoms N1 and N2, then the interfragmental bond is the one between
N1 and N2. If the user does not define the interfragmental bond, it is constructed between the atom pair
with nearest distance between the two fragments. Then the angles and dihedrals are constructed upon this
(different) set of bonds in the already described fashion.
Now let us regard the definition of the fragment constraints: A fragment is constrained internally by
constraining all internal coordinates that contain only atoms of the respective fragment. The connection
between two fragments A and B is constrained by constraining specific internal coordinates that contain
atoms of both fragments. For bonds, one atom has to belong to fragment A and the other atom has to belong
to fragment B. Regarding angles, two atoms have to belong to fragment A and one to fragment B and vice
versa. With respect to dihedrals, only those are constrained where two atoms belong to fragment A and the
other two belong to fragment B.
9.17.2 Transition State Optimization
As transition state finder we implemented the well-established eigenvector following algorithm using a P-RFO
step as implemented by Baker [129]. This algorithm is a quasi-Newton like algorithm.
The Taylor series of the energy, truncated after the quadratic term, is:
1
E = E0 + gq + qq + q + Hq q (9.225)
2
The Newton-Raphson step to get from the actual point to a stationary point is:
X Vi+ gq Vi
q = H1
q gq = (9.226)
bi
with Vi and bi as eigenvectors and eigenvalues of the Hessian Hq . This step leads to the nearest stationary
point on the PES. This stationary point can be a minimum or a saddle point, according to the curvature of
the PES at the actual point.
With a simple shift of the Hessian eigenvalues bi in this equation one can guide the step to a stationary point
with the required characteristics (Hessian with exactly one negative eigenvalue). The transition state search
is separated into two different optimization problems. The energy is maximized along one Hessian eigenmode
and minimized along the remaining 3N 7(6) eigenmodes. We introduce two different shift parameters
p and n , where p is the shift parameter for the eigenmode being maximized and n shifts the Hessian
eigenvalues of the modes being minimized. This method allows us to maximize along any mode, not only
the one with smallest eigenvalue. Starting from two different RFO-matrices for the different optimization
problems (see description above) we get for p and n :
1 1 Fi2
q X
p = bk + b2k + 4Fk2 and = n (9.227)
2 2 n bi
i6=k
whereas Fi = Vi+ g is the component of g along the Hessian eigenmode Vi and n has to get solved iteratively.
The solution for n has to be negative and lower than b2 (or lower than b1 , if not the lowest mode is being
followed). If the Hessian has more than one negative eigenvalue, these properties might not be fulfilled, and
the Hessian would have to be modified. In our implementation the Hessian diagonal elements are either
shifted or reversed in such a case.
Once the shift parameters are known the P-RFO step h is calculated as follows:
Fk Vk Fi Vi
qk = and qi = with i = 1 . . . n, i 6= k (9.228)
bk p bi n
n
X
q = qj (9.229)
j=1
ScanTS option
For TS modes of rather local nature (involving only one bond or an angle; no concerted movements over
multiple atoms) we implemented the ScanTS feature. Here the user can carry out a relaxed surface scan and
a TS optimization in one calculation. After the relaxed surface scan the algorithm chooses the optimized
structure of the scan with highest energy as initial guess structure and the two neighbouring structures for the
calculation of the second derivative of the scanned coordinate (e.g., if scan step number 4 gives the structure
with highest energy, then structure basename.004.xyz is the initial guess for the TS optimization; the
structures basename.003.xyz and basename.005.xyz are used for the calculation of the second derivative).
Before the first step of the subsequent TS optimization the energies and gradients for all three structures are
calculated. The gradients are then transformed to internal coordinates. The diagonal Hessian value of the
scanned coordinate is then calculated via finite difference of the internal gradients of the two given structures
(003 and 005 in our example).
For the construction of the initial Hessian a model force field Hessian is built up (this Hessian has got only
diagonal entries and zeros as off-diagonal elements). The exactly calculated diagonal Hessian value replaces
the model force field Hessian entry for the respective internal coordinate.
If the user already performed a regular relaxed surface scan without the subsequent TS optimization, then
he can nevertheless use these structures for the same procedure. A relaxed surface scan always gives you
the xyz-files and gbw-files of the optimized structures of each scan step. A separate TS optimization can
be carried out where the structure with highest energy is the starting structure. Additionally the two files
with the two adjacent structures (as explained above) have to be provided (via the Hess Internal keyword,
see below). Furthermore, the internal coordinate, for which the diagonal Hessian value has to be calculated,
has to be given (the previously scanned coordinate). This exact Hessian calculation is only possible for one
internal coordinate:
%geom
Hess_Internal
{B 1 0 C} # previously scanned coordinate
XYZ1 "scanName.003.xyz" # the xyz-files of the structures
XYZ2 "ScanName.005.xyz" # next to the highest energy point
GBW1 "ScanName.003.gbw" # the gbw-files of the structures
GBW2 "ScanName.005.xyz" # next to the highest energy
# the gbw-files are optional
end
end
Additionally the manipulation of the diagonal Hessian values of the internal Hessian is possible for further
internal coordinates, but without an extra calculation. Here the user can just define a value (in Eh/Bohr2 ).
Hess_Internal
{A 3 2 1 D 2.0} # define a diagonal Hessian value of
# 2 Eh/Bohr2 for the angle between
# atoms 3 2 1
{B 1 0 D -0.5} # define a diagonal Hessian value of
# -0.5 Eh/Bohr2 for the bond between
# atoms 1 and 0
end
The definition of such Hessian (diagonal) elements is possible for multiple internal coordinates. These just
replace the values of the force field model Hessian.
Hybrid Hessian
We implemented the calculation of a Hybrid Hessian as an alternative to the full Hessian calculation for
TS optimization. Here only those parts of the Hessian, that are important for the TS optimization, are
calculated exactly. For this calculation we define two kinds of atoms: atoms whose couplings with the other
atoms are treated exactly (E) and atoms whose couplings are treated approximately (A).
In a first step an Almloef model Hessian is built up in redundant internal coordinates and transformed to
Cartesian coordinates. This Hessian gives the second derivative elements for atom pairs A/A. In a second
step the second derivative elements between pairs E/E and E/A are calculated numerically as in a numerical
frequency calculation:
i,B eq.
E E gj,C gj,C
= = (9.230)
iB jC jC iB displ.
with:
i, j x-, y- or z-direction
B, C pairs of E/E, E/A, A/E
displ. magnitude of displacement
eq.
gj,C force on atom C in direction j in current geometry
i,B
gj,C force on atom C in direction j after displacement of atom B in direction i
Partial Hessian Vibrational Analysis
We implemented the Partial Hessian Vibrational Analysis (PHVA), as published by Li, Jensen in [132], for
the analysis of the nature of stationary points of structures obtained with QM/MM optimizations.
# PHVA after a QM/MM optimization in the (dispersion-/PC-) field

# caused by the MM-atoms
! NumFreq
%LJCoefficients "temp.LJ" # file with the Lennard Jones
# coefficients for dispersion interaction
# obtained from last QM/MM run
%pointcharges "temp.pc" # file with the point charges for
# electrostatic interaction
# obtained from last QM/MM run
#
%freq
PARTIAL_Hess {0 1 2} # atoms which are "frozen" and which make
# the boundary to the MM-system
end
end
NOTE
This procedure should be used for QM/MM optimized structures only to verify the nature of the
stationary point and have an estimate of the ZPE.
Here we shortly describe the procedure: In PHVA we divide the system into two parts B (of size n atoms)
and C (size N n). Let the atom set B belong to the region where the chemical change is localized. The
Partial Hessian matrix is built up as follows:
!
KBB 0

(9.231)
0 KCC
With:
KBB : x, y, z direction (9.232)

0 0

KCC

= .. 8
0 . 0 , = 10 au, (9.233)

0 0
this corresponds to using near-infinite masses for the atoms in C.
With this procedure we get the following eigenvalue structure:
Six zero eigenvalues with modes corresponding to translational and rotational motion of the entire
molecule.
3(N n) 6 small (less than 1 cm1 ) eigenvalues with modes corresponding mainly to internal motion
within region C.
Three eigenvalues (typically less than 10 cm1 ) with modes corresponding mainly to motion of region
C relative to region B.
(3n 3) eigenvalues with modes corresponding mainly to relative motion of B and C as well as internal
motion within region B.
9.17.3 Minimum Energy Crossing Points
The MECP optimization allows the user to optimize to structures where two different potential energy
surfaces (PES1 and PES2) cross each other. In this optimization two conditions apply: the energy E1 of
2
PES1 is minimized while at the same time the energy difference (E1 E2 ) of both surfaces is minimized.
For the implementation we follow in principle the suggestions of Harvey et al. in [131].
For the minimization two different gradients are constructed:
The first gradient chosen for the minimization is
2
f= (E1 E2 ) = 2 (E1 E2 ) x1 (9.234)
q
where x1 is the gradient difference vector

E1 E2
x1 = (9.235)
q q
which is orthogonal to the crossing hyperline near the MECP.
The gradient

E1 x1 E1 x1
g= (9.236)
q |x1 | q |x1 |
is orthogonal to x1 .
Both gradients are combined to yield the effective surface crossing gradient
gSC = g + f (9.237)
The crossing hyperline is defined as the 3N 7 dimensional subspace of PES1, which is orthogonal to x1 . In
the MECP optimization we want to find the point of lowest energy within this subspace.
Our calculation of normal modes and force constants for movements along the crossing hyperline differ
from the one proposed by Harvey et al. A standard frequency analysis can not be performed, but a similar
procedure is applied:
Let us regard the second-order Taylor expansion for the energy of both surfaces near the MECP for a
displacement along the crossing hyperline (orthogonal to x1 ):
1
EA = EMECP + q T Heff,A q (9.238)
2
with:
EA Energy E1 on PES1 or E2 on PES2

Heff,A effective Hessian for PES1 or PES2
q displacement along the crossing hyperline
Diagonalization of this effective Hessian gives us the normal modes of the crossing hyperline and thus allows
us to decide whether the MECP optimization converged to a minimum in the 3N 7 dimensional subspace
of the crossing hyperline.
The procedure for the calculation of the effective Hessian is now as follows: For each of both surfaces the
second derivative matrix is calculated. Then the 6 rotations and translations and additionally the direction of
the gradient difference vector x1 (this ensures that movement orthogonal to the crossing hyperline, for which
we do NOT satisfy the conditions of a stationary point, is excluded) are projected out from the Hessian
matrix.
For MECP optimizations the following options exist:
%mecp
SurfCrossOpt true # switches on the MECP optimization
# alternatively use: ! SurfCrossOpt
SurfCrossNumFreq true # switches on the MECP effective Hessian
# calculation
# alternatively use: ! SurfCrossNumFreq
# separate MO input for the second spin state (PES2)
moinp "Myfile.gbw"# MO input for PES2
# information on the electronic structure of PES 2
Mult 3 # multiplicity of PES2
brokenSym 1,1 # broken symmetry for PES2
# CASSCF options for PES2 (also see the CASSCF chapter)
casscf_nel 6 # number of active space electrons
casscf_norb 6 # number of active orbitals
casscf_mult 1,3 # multiplicities singlet and triplet
casscf_nroots 4,2 # four singlets, two triplets
casscf_bweight 2,1 # singlets and triplets weighted 2:1
casscf_weights[0] = 0.5,0.2,0.2,0.2 # singlet weights
casscf_weights[1] = 0.7,0.3 # triplet weights
end
9.17.4 Numerical Gradients
If you want to use numerical instead of analytic gradients you have to use
! NumGrad
in your input file. Additionally the settings for the numerical differentiation can be changed:
%numgrad
CentralDiff true # (Default) You should use two-sided numerical
# differentiation, but it is possible to switch
# to one-sided numerical differentiation.
DX 0.005 # Increment in Bohr for the differentiation.
TransInvar true # (Default) Take advantage of translation invariance
end
9.17.5 ORCA as External Optimizer
If you want to use only the optimizer of ORCA you have to use
! ExtOpt
in your input file. All information that you give on the electronic structure is discarded. In each optimization
step ORCA writes a file called extopt.extcomp.inp with the following format:
"N (nr. of atoms)" "calculation requested" "charge" "multiplicity"

atomic nr.(atom1) x-coord.(atom1) y-coord.(atom1) z-coord.(atom1)
...
atomic nr.(atomN) x-coord.(atomN) y-coord.(atomN) z-coord.(atomN)
example:
9 1 0 1
6 -7.183647732744 1.832728827744 -0.116462028781
9 -6.344271117689 -0.569555640677 -0.053332293594
9 -6.385938620292 3.010680341363 1.994410067976
9 -9.725973988287 1.760800299423 -0.041860336809
6 -6.281650723853 3.228508119832 -2.541654671259
1 -7.049215894384 2.192752895053 -4.164371259742
1 -7.120030511673 5.123374809616 -2.502306840221
8 -3.579612411580 3.262825146858 -2.485537715188
1 -3.134995715660 4.196025355887 -4.047828323839
NOTE: the coordinates are given in Bohr.
There are two types of calculation:
1. calculate energy and gradient
2. calculate energy, gradient and Hessian
If a Hessian is requested, it should be stored in the ORCA Hessian file format and be named yourInputFilename.hess.
ORCA then calls a script (that is not distributed with the ORCA binaries):
orca External extopt.extcomp.inp extopt.extcomp.out
Your script starts the energy, gradient (and Hessian) calculation and finally provides the results in a file
called extopt.extcomp.out, which has to be written in the following format:
Total Energy
gradient(x) on atom1 gradient(y) on atom1 gradient(z) on atom1
...
gradient(x) on atomN gradient(y) on atomN gradient(z) on atomN
example:
-1.135276876846e+02
-1.144030900000e-05 3.458417100000e-05 7.686904800000e-06
-4.450221700000e-05 -2.016015000000e-05 -2.617359400000e-05
4.460242300000e-05 -3.290908700000e-05 5.698279500000e-06
1.026743300000e-06 4.889225700000e-05 3.474765100000e-05
6.178236500000e-05 -1.506876000000e-04 -1.288283000000e-04
1.601578300000e-05 1.670611600000e-05 2.747591400000e-05
-4.360515500000e-06 5.840020000000e-06 8.823311100000e-06
-1.135615900000e-04 1.384714300000e-04 6.197283200000e-05
5.043762200000e-05 -4.073673300000e-05 8.597172100000e-06
NOTE:
the energy should be given in Eh, gradients should be given in Eh/Bohr, Hessian values should be
given in Eh/Bohr2
ORCA then performs the next optimization step, checks for convergence, and if the optimization has not yet
converged, it goes on to the next optimization step.
Gaussian as External Optimizer. To use the external optimizer from Gaussian in ORCA, the following
keywords were provided in the past:
9.18 Excited States via RPA, CIS and TD-DFT 495
%geom
UseGaussian true # Use the external Gaussian optimizer instead
# of the ORCA optimizer.
GaussianName "GAU" # String defining the name of the Gaussian
# optimizer
GauOptFlags # String indicating the optimization flags
Gaussian Constraints # List defining the constraints for
# the Gaussian optimizer.
end
Since the ORCA team got banned by Gaussian in January 2007 we can no longer support these option flags.
They have not been removed from the code and may or may not work. If there is trouble with it we can
unfortunately not offer any help since we do not have access to the Gaussian code any longer.
9.18 Excited States via RPA, CIS and TD-DFT
ORCA features a relatively efficient single-excitation CI (CIS), random-phase approximation and time-
dependent DFT module that can be used to calculate excitation energies, absorption intensities and CD
intensities. Especially TD-DFT became very popular for excited state calculations as it offers significantly
better results than HF-CIS at about the same cost. However, there are also many pitfalls of TD-DFT,
some of which are discussed in reviews [373] [374]. TD-DFT methods are available for closed-shell and
spin-unrestricted reference states. Analytic gradients are available. There also is a doubles correction
implemented that improves the results (but also the computational cost). It is often used together with
double-hybrid functionals as explained below. The TD-DFT module of ORCA is also extensively used for the
calculation of X-ray absorption spectra at the K-edge of a given element.
9.18.1 General Features
The module is invoked with the block:
%cis end
# or equivalently
%tddft end
There are a variety of options. The most important one is the number of excited states that you want to
have calculated:
%cis NRoots 10
end
The variable NRoots gives the number of excited states to be calculated. The expansion space in the Davidson
procedure is limited by MaxDim NRoots. With MaxDim values of 5-10 the calculations will show favorable
convergence but also increased disk space demands. In general the larger NRoots the more core memory is
needed while the disk space requirements are proportional to MaxDim.
The convergence tolerances are:
%cis ETol 1e-6

RTol 1e-6
end
The variable ETol gives the required convergence of the energies of the excited states (in Eh) and RTol is the
required convergence on the norm of the residual vectors. Under normal ciorcumstances the calculations need
about 5-10 iterations to converge to the default convergence tolerances.
If closed-shell references are used the program can calculate the singlet and triplet excited states at the same
time by using:
%cis triplets true

end
This is available for all combinations of methods except double-hybrid DFT or analytic gradients.
In order to control the orbitals that should be taken into account in the calculation two mechanisms are
available. The first mechanism is the default mechanism and consists of specifying and orbital energy window
within which all single excitations will be considered:
%cis EWin -3,3 # (orbital energy window in Eh)

end
Thus, the default is to keep core orbitals frozen and to neglect very high lying virtual orbitals which is a
sensible approximation. However, you may want to consider to include all virtual orbitals by choosing for
example EWin -3,10000. The second mechanism is to explicitly give an orbital energy window for each
operator, i.e.
%cis OrbWin[0] 2,-1,-1,14 # orbital window for spin-up MOs

OrbWin[1] 2,-1,-1,16 # orbital window for spin-down MOs
end
The -1s in the above example mean that the HOMO and LUMO for the spin-.up and spin-down orbitals
will be automatically determined by the program.
In using the CIS/TD-DFT module five different types of calculations should be distinguished:
Semiempirical methods
Hartree-Fock calculations
DFT calculations without HF exchange (non-hybrid functionals)
DFT calculations with HF exchange (hybrid functionals)
DFT calculations with HF exchange and MP2 correlation (double-hybrid functionals)
9.18.2 Semiempirical Methods
The semiempirical INDO/S method is very suitable to calculate absorption spectra of medium sized to large
organic and inorganic molecules. It has been parameterized by the late M. C. Zerner for optical spectroscopy
and in my experience at least, it tends to work nicely for many systems. With the semiempirical approach
it is easy to calculate many states of large molecules. For example, consider the following calculation on a
bis-histidine ligated iron-porphyrin model (in the Fe(II) state) that includes 92 atoms and 16,500 CSFs in
the single excitation space. Yet the calculation requires only a few minutes on an ordinary computer for the
prediction of the first 40 excited states.
The calculated spectrum is in essentially reasonable agreement with experiment in showing a huge band
around 400 nm (the famous Soret band) and a smaller but still intense band between 500 and 550 nm (the
Q-band). There are no predicted absorptions below 10,000 cm1 .
The input for the job is shown below:
# Test CIS in conjunction with INDO/S
! RHF ZINDO/S TightSCF DIIS NoRICO NoMOPrint

%cis NRoots 40
end
* xyz 0 1
Fe -0.01736 0.71832 -0.30714
C 2.65779 4.03195 -0.13175
C 3.51572 3.02488 -0.24101
C 2.66971 1.82027 -0.30891
C 3.30062 0.51609 -0.42755
C 2.61022 -0.60434 -0.47131
C 3.32146 -1.89491 -0.57434
C 2.35504 -2.79836 -0.57179
C 1.11740 -1.99868 -0.46878
C -0.04908 -2.61205 -0.44672
C -1.30967 -1.89127 -0.38984
C -2.58423 -2.63345 -0.40868
C -3.50492 -1.68283 -0.37930
C -2.72946 -0.42418 -0.33711
C -3.35747 0.73319 -0.28970
C -2.66935 2.01561 -0.22869
C -3.31167 3.19745 -0.16277
C -4.72835 3.62642 -0.14517
C -5.84825 2.89828 -0.20597

C -2.21443 4.15731 -0.09763
C -1.11572 3.39398 -0.14235
C 0.19578 4.02696 -0.10122
C 1.33370 3.36290 -0.15370
C 3.09165 5.44413 -0.02579
C 2.35656 6.55323 0.10940
N 1.43216 2.09428 -0.24815
N 1.34670 -0.74673 -0.42368
N -1.39885 2.15649 -0.21891
N -1.47620 -0.63353 -0.34705
C 5.03025 3.02708 -0.28544
C 4.81527 -2.12157 -0.66646
C -5.01065 -1.83771 -0.38886
C -2.28137 5.66820 -0.00321
C -2.73691 -4.14249 -0.43699
C -2.42579 -4.72805 -1.83259
C 2.45978 -4.31073 -0.64869
C 2.19678 -4.82182 -2.08201
C 1.60835 -6.22722 -2.10748
C -1.90102 -6.15737 -1.82447
O -1.96736 -6.92519 -2.75599
O 1.60982 -7.01844 -1.19330
O -1.15355 -6.41323 -0.74427
O 0.89871 -6.41433 -3.22828
H 4.17823 5.62170 -0.05623
H 2.86221 7.53117 0.17503
H 1.26303 6.57673 0.17212
H 0.21799 5.11603 -0.03468
H -1.78003 6.14426 -0.87498
H -3.32281 6.05139 0.01906
H -1.78374 6.03115 0.92347
H -4.89690 4.71221 -0.07358
H -6.82566 3.40843 -0.18007
H -5.88239 1.80643 -0.28628
H -4.44893 0.70720 -0.28575
H -5.32107 -2.89387 -0.54251
H -5.45075 -1.49552 0.57400
H -5.46788 -1.24144 -1.20929
H -2.05997 -4.55939 0.34045
H -3.76430 -4.43895 -0.12880
H -3.33638 -4.66246 -2.47119
H -1.65517 -4.10119 -2.33605
H -0.56422 -7.14866 -1.00437
H 0.26056 -7.12181 -3.00953
H 1.48118 -4.13253 -2.58671
H 3.13949 -4.79028 -2.67491
H 3.46153 -4.65168 -0.30336
H 1.73023 -4.75206 0.06633

H 5.26172 -1.51540 -1.48550
H 5.31767 -1.84036 0.28550
H 5.06416 -3.18438 -0.87628
H -0.07991 -3.70928 -0.48866
H 4.39835 0.46775 -0.47078
H 5.39550 2.59422 -1.24309
H 5.47197 4.04179 -0.19892
H 5.44914 2.41988 0.54738
N 0.01831 0.60829 1.68951
C 0.02054 1.64472 2.54371
C 0.04593 -0.50152 2.45186
N 0.04934 1.20474 3.84418
C 0.06582 -0.16578 3.80848
H 0.00322 2.72212 2.31829
N -0.05051 0.81937 -2.30431
H 0.05251 -1.53704 2.08183
C 0.11803 1.92670 -3.04495
H 0.05712 1.81091 4.70485
H 0.08982 -0.83278 4.68627
C -0.24302 -0.18840 -3.17641
C -0.19749 0.28568 -4.49059
N 0.03407 1.63309 -4.38373
H 0.30109 2.95786 -2.70479
H -0.41432 -1.24242 -2.91290
H -0.31761 -0.27403 -5.43315
H 0.12975 2.31943 -5.17616
*
Figure 9.19: Structure of the iron-porphyrin used for the prediction of its absorption spectrum (the
structure was obtained from a molecular mechanics calculation and the iron-imidazole
bondlength was set to 2.0 A).
Figure 9.20: The ZINDO/S predicted absorption spectrum of the model iron porphyrin shown
above. The spectrum has been plotted using the orca mapspc tool.
Note that ORCA slightly departs from standard ZINDO/S in using dipole integrals in the intensity calculations
that include all one- and two-center terms which are calculated via a STO-3G expansion of the Slater basis
orbitals. The calculated intensities are not highly accurate anyways. In the present case they are overestimated
by a factor of 2.
9.18.3 Hartree-Fock Wavefunctions
When applying the procedures outlined above to pure Hartree-Fock, one obtains the random-phase approxi-
mation (RPA) or the CI singles (CIS) model (when effectively using the Tamm-Dancoff Approximation,
TDA). In general, RPA and CIS calculations do not lead to good agreement with experimental excitation
energies and errors of 1-5 eV are common. Therefore HF/CIS is mostly a qualitative tool or can be used with
caution for larger molecules if more extensive and more well balanced CI calculations are not computationally
tractable.
9.18.4 Non-Hybrid and Hybrid DFT
For DFT functionals there is the choice between the full TD-DFT (eq. 9.239) treatment and the so-called
Tamm-Dancoff approximation (TDA).
! ! ! !
A B X 0 X
= (9.239)
B* A* Y 0 Y
The TDA is the same approximation that leads from RPA to CIS (i.e. neglect of the so-called B matrix, see
eq. 9.240). The results for vertical excitation energies are usually very similar between the two approaches.
AXTDA = TDA XTDA (9.240)
In general, the elements of matrix A and B for singlet-singlet excitations in the spin-restricted case are
given by eqs. 9.241 and 9.242.
Aia,jb = ij ab (a i ) + 2(ia|jb) aX (ij|ab)

(9.241)
+ (1 aX )(ia|fXC |jb)
and
Bia,jb = 2(ia|bj) aX (ib|aj) + (1 ax )(ia|fXC |bj). (9.242)
Here, i, j denote occupied and a, b virtual orbitals. aX is the amount of non-local Fock exchange in the
density functional. If aX is equal to one, eqs. 9.239 and 9.240 correspond to the RPA and CIS case, based on
a Hartree-Fock ground state determinant.
The TDA is turned on by:
%tddft TDA true # (default)

TammDancoff true # (equivalent)
end
There are situations where hybrid functionals give significantly better results than pure functionals since they
suffer less from the self-interaction error. The RIJCOSX procedure [104] [248] leads to very large speedups in
such calculations at virtually no loss in accuracy [375].
9.18.5 Simplified TDA and TD-DFT
ORCA also supports calculations of excited states using the simplified Tamm-Dancoff approach (sTDA) by
S. Grimme [376]. The sTDA is particularly suited to calculate absorption spectra of very large systems.
sTDA as well as the simplified time-dependent density functional theory (sTD-DFT) [377] approach require
a (hybrid) DFT ground state calculation. For large systems, using range-separated hybrid functionals (e.g.
B97X) is recommended. [378]
The sTD-DFT approach in particular yields much better electronic circular dichroism (ECD) spectra and
should be used for this purpose.
9.18.5.1 Theoretical Background
A brief outline of the theory will be given in the following. For more details, please refer to the original
papers [376, 377]. In the sTDA, the TDA eigenvalue problem from eq. 9.240 is solved using a truncated and
semi-empirically simplified A0 matrix. The trunctation negelects all excitations that are beyond the energy
range of interest, except a few strongly coupled ones. The matrix elements from eq. 9.241 are simplified by
neglecting the response of the density functional and by approximating the remaining two-electron integrals
as damped Coulomb interactions between transition/charge density monopoles. In the following, the indices
i, j denote occupied, a, b virtual and p, q either kind of orbitals.
NX
atoms
A0ia,jb = ij ab (a i ) + A K B
(2qia A J
AB qjb qij B
AB qab ) (9.243)
A,B
A B
qpq and qpq are the transition/charge density monopoles located on atom A and B, respectively. These are
obtained from Lowdin population analysis (see Sec. 9.30.3). p is the Kohn-Sham orbital energy of orbital p.
K J
AB and AB are the Mataga-Nishimoto-Ohno-Klopman damped Coulomb operators for exchange-type (K)
and Coulomb-type (J) integrals, respectively.
1
J 1
AB = (9.244)
(RAB ) + (aX )
1
K 1
AB = (9.245)
(RAB ) +
Here, is the arithmetic mean of the chemical hardness of atom A and B. and are the parameters of the
method and are given by:
= 1 + ax 2 (9.246)
= 1 + ax 2 (9.247)
For any global hybrid functional, 1 , 2 , 1 and 2 are identical. and then depend on the amount of
Fock exchange (aX ) only. This is different for range-separated hybrid functionals where 2 and 2 are set to
zero. 1 and 1 along with a value ax for the sTDA treatment are individually fitted for each range-separated
hybrid functional. [378] It can bee seen from eq. 9.243 that the method is asymptotically correct which is
crucial for excitations of charge transfer type.
In sTD-DFT, eq. 9.239 is solved using the simplified matrices A0 (see above) and B 0 .
NX
atoms
0 A K B A K B
Bia,jb = (2qia AB qbj aX qib AB qaj ) (9.248)
A,B
This approach yields better transition dipole moments and therefore spectra but the method is more costly
than sTDA (a factor of 25 for typical systems). The parameters used in sTDA and sTD-DFT are identical.
There are no additional parameters fitted for this method.
9.18.5.2 Calculation Set-up
sTDA and sTD-DFT can be combined with any (restricted or unrestricted) hybrid DFT singlepoint calculation.
Gradients and frequencies are not implemented! The methods can be invoked via the %tddft block. Table
9.13 gives a list of the possible keywords.
Table 9.13: Keyword list for sTDA and sTD-DFT.

Mode sTDA Invokes a sTDA calculation
Mode sTDDFT Invokes a sTD-DFT calculation
EThresh value Energy threshold up to which CSFs are included (in eV)
PTLimit value Energy threshold up to which CSFs beyond EThresh may be selected (in eV)
PThresh value Selection criterion to include CSF beyond EThresh (in Eh)
axstda value Fock exchange parameter used in sTDA/sTD-DFT calculation (for range-separated hybrids)
beta1 value Constant part of J integral parameter
beta2 value aX scaled part of J integral parameter
alpha1 value Constant part of K integral parameter
alpha2 value aX scaled part of K integral parameter
triplets true Calculate singlet-triplet excitations (default: singlet-singlet)
The following example shows how to run such a sTDA calculation using the BHLYP functional if one is
interested in all excitations up to 10 eV.
! bhlyp def2-SV(P) rijcosx gridx5 nososcf tightscf

! smallprint printgap nopop
%maxcore 5000
%tddft
Mode sTDA
Ethresh 10.0
maxcore 5000
end
* xyzfile 0 1 coord.xyz
Replacing Mode sTDA by Mode sTDDFT will invoke a sTD-DFT calculation instead. This is shown in the next
example in combination with the B97X functional and user specified parameters:
! wb97x def2-SV(P) rijcosx gridx5 nososcf tightscf

! smallprint printgap nopop
%maxcore 5000
%tddft
Mode sTDDFT
Ethresh 10.0
axstda 0.56
beta1 8.00
beta2 0.00
alpha1 4.58
alpha2 0.00
maxcore 5000
end
For the range-separated hybrid functionals LC-BLYP, CAM-B3LYP, B97, B97X and B97X-D3, parameters
are available and will be used by default if one of these functionals is used. The way of specifying parameters
as shown above is useful if there is a range-separated hybrid functional that has not been parametrized for
sTDA yet. For very large systems (e.g. > 500 atoms), it may be useful to define an upper boundary PTLimit
for the selection of configurations that are beyond EThresh (otherwise the whole configuration space will be
scanned). This can be done as shown below:
! cam-b3lyp grid5 nofinalgrid def2-SV(P) nori tightscf

! nososcf smallprint printgap nopop
%pal nprocs 4
end
%maxcore 5000
%tddft
Mode sTDDFT
Ethresh 10.0
PThresh 1e-4
PTLimit 30
maxcore 20000
end
%method
runtyp energy
end
In this case, all excitations up to 7 eV are considered from the very beginning. Configurations between 7
and 14 eV are included if their coupling to the configurations below 7 eV is strong enough (in total larger
than PThresh). All configurations beyond 14 eV are neglected. Since the sTDA/sTD-DFT calculations
run in serial mode, it is recommended to reset the maxcore within the %tddft block (as done in the above
examples). In the latter sample input, the ground state procedure runs in parallel mode on 4 cores with a
maxcore of 5000 MB set for each node. The subsequent sTD-DFT calculation then runs on a single core,
but in order to use all the available memory, the maxcore is reset to a larger value (i.e., 20000 MB). If the
maxcore statement within the %tddft block was missing, only 5000 MB of memory would be available in
the sTD-DFT calculation. Note furthermore that for very large systems, using a functional with the correct
asymptotic behaviour is very important (due to the fixed amount of GGA exchange, CAM-B3LYP does not
provide this property).
The ORCA output will summarize the important properties of your calculation which allows you to check
your input:
---------------------------------------------------------------------------------
ORCA sTDA CALCULATION
please cite in your paper

orginal sTDA method: S. Grimme, J. Chem. Phys. 138, 244104 (2013)
range-separated sTDA: T. Risthaus, A. Hansen, S. Grimme, Phys. Chem. Chem. Phys.
16, 14408-14419 (2014)
sTD-DFT approach: C. Bannwarth, S. Grimme, Comp. Theor. Chem.
1040-1041, 45-53 (2014)
---------------------------------------------------------------------------------
spectral range up to (eV) ... 10.000000

occ. MO cut-off (eV) ... -24.052589
virt. MO cut-off (eV) ... 17.726088
perturbation threshold ... 1.000e-04
CSF selection range up to (eV) ... 30.000000
MOs in sTD-DFT ... 37
occ. MOs in sTD-DFT ... 14
virt. in sTD-DFT ... 23
calculate triplets ... no
Calculating the dipole lengths integrals ...

Transforming integrals ...
Calculating the dipole velocity integrals ...
Calculating magnetic dipole integrals ...
SCF atom population (using active MOs):
4.009 4.182 4.182 4.318 4.318 0.867 0.867 0.876 0.876 0.876
0.876 0.876 0.876
Number of electrons in sTDA: 28.000
ax(DF) : 0.3800
s_k : 2.0000
beta (J): 1.8600
alpha (K): 0.9000
The spectroscopic data is also printed out after the calculation has finished:
14 roots found, lowest/highest eigenvalue : 6.627 9.945
excitation energies, transition moments and amplitudes
molecular weight: 68.119

state eV nm fL fV Rl RV
0 6.627 187.1 0.000000 0.000001 0.002400 0.033014 0.71 ( 12-> 14) ...
1 6.637 186.8 0.000188 0.000233 -6.595360 -6.544674 -0.71 ( 13-> 14) ...
2 8.162 151.9 0.000022 0.000113 -0.169704 -0.383021 -0.65 ( 12-> 16) ...
3 8.185 151.5 0.708166 0.559459 -33.378989 -33.157817 0.62 ( 13-> 16) ...
4 8.514 145.6 0.461396 0.349012 64.100474 55.364958 -0.63 ( 12-> 17) ...
5 8.531 145.3 0.000004 0.000282 0.539213 4.637973 -0.72 ( 13-> 17) ...
6 8.927 138.9 0.000080 0.001340 0.439265 1.794914 0.70 ( 13-> 18) ...
7 8.929 138.9 0.002612 0.003077 -5.590091 -7.144206 -0.69 ( 12-> 18) ...
8 9.156 135.4 0.432008 0.300685 -30.271745 -29.351033 -0.74 ( 12-> 17) ...
9 9.347 132.6 0.058500 0.054136 -37.502752 -36.077121 -0.53 ( 12-> 19) ...
10 9.534 130.0 0.338851 0.235400 59.709273 68.042758 0.66 ( 12-> 18) ...
11 9.624 128.8 0.007213 0.004968 25.554619 21.208832 -0.49 ( 13-> 18) ...
12 9.922 125.0 0.021172 0.019486 -22.874039 -23.258574 0.81 ( 13-> 20) ...
13 9.945 124.7 0.001403 0.001498 6.301469 6.510456 0.79 ( 12-> 20) ...
sTD-DFT done
Total run time: 0.326 sec
*** ORCA-CIS/TD-DFT FINISHED WITHOUT ERROR ***
fL, fV, RL and RV are the length and velocity expressions of the oscillator and rotatory strengths, respectively.
They may be convoluted by a spectrum processing program to yield the UV/Vis absorption and ECD
spectra.
9.18.6 Double-hybrid functionals and Doubles Correction
The program can compute a doubles correction to the CIS excitation energies. The theory is due to Head-
Gordon and co-workers. [379] The basic idea is to compute a perturbative estimate (inspired by EOM-CCSD
theory) to the CIS excited states that is compatible with the MP2 ground state energy. In many cases this is
a significant improvement over CIS itself and comes at a reasonable cost since the correction is computed a
posteriori. Of course, if the CIS prediction of the excited state is poor, the (D) correction being perturbative
in nature cannot compensate for qualitatively wrong excited state wavefunctions.
In addition and perhaps more importantly the (D) correction is compatible with the philosophy of the
double-hybrid functionals and should be used if excited states are to be computed with these functionals.
The results are usually much better than those from TD-DFT since due to the large fraction HF exchange,
the self-interaction error is much smaller than for other functionals and after the (D) correction the results
do not suffer from the overestimation of transition energies that usually comes with increased amounts of HF
exchange in TD-DFT calculations.
Since the calculations would require a fairly substantial integral transformation that would limit it to fairly
small molecules if no approximation are introduced we have decided to only implement a RI version of it.
With this approximation systems with more than 1000 basis functions are readily within the reach of the
implementation.
Since one always has a triad of computational steps: MP2-CIS solution-(D) correction, we have implemented
several algorithms that may each become the method of choice under certain circumstances. The choice
depends on the size of the system, the number of roots, the available main memory and the available disk
space together with the I/O rate of the system. The formal cost of the (D) correction is O(N 5 ) and its
prefactor is higher than that of RI-MP2. In the best case scenario, the rate limiting step would be the
calculation of the pair-contribution in the U-term which requires (for a closed-shell system) twice the effort
of a RI-MP2 calculation per state.
The use of the (D)-correction is simple. Simply write:
! RKS B3LYP/G SV(P) SV/C TightSCF

%cis dcorr n # n=1-4. The meaning of the four algorithms is
# explained below.
# algorithm 1 Is perhaps the best for small systems. May use a
# lot of disk space
# algorithm 2 Stores less integrals
# algorithm 3 Is good if the system is large and only a few
# states are calculated. Saves disk and main
# memory.
# algorithm 4 Uses only transformed RI integrals. May be the
# fastest for large systems and a larger number
# of states
end
Table 9.14: Integral handling in various implementations of the (D) correction (i,j=occupied MOs, a,b=virtual
MOs, Q=aux function; NumInt=numerical integration).
DCORR= 1 2 3 4
(ia|jb) integrals Stored Stored Not stored Not stored
(ij|ab) integrals Stored Not made Not made Not made
(ab|Q) integrals Stored Not made Not made Stored
(ij|Q) integrals Stored Stored Stored Stored
(ia|Q) integrals Stored Stored Stored Stored
Coulomb CIS From (ia|jb) From (ia|jb) From (ia|Q) From (ia|Q)
Exchange CIS From (ij|ab) RI-AO-direct RI-AO-direct From (ab|Q)
XC-CIS Num. Int. Num. Int. Num. Int. Num. Int.
V-term in (D) From (ia|jb) From (ia|jb) From (ia|Q) From (ia|Q)
U-term in (D) From (ab|Q) RI-AO-direct RI-AO-direct From (ab|Q)
NOTE:
In all three involved code sections (MP2, CIS, (D)) the storage format FLOAT is respected. It cuts
down use of disk and main memory by a factor of two compared the default double precision version.
The loss of accuracy should be negligible; however it is as always in science better to double check.
The (ab|Q) list of integrals may be the largest for many systems and easily occupies several GB of disk
space (hence algorithms 2 and 3). However, that disk-space is often well invested unless you run into
I/O bottlenecks.
The (ia|jb) and (ij|ab) lists of integrals is also quite large but is relatively efficiently handled. Nevertheless,
I/O may be a problem.
Making the exchange contribution to the CIS residual vector in an RI-AO direct fashion becomes quite
expensive for a larger number of states. It may be a good choice if only one or two excited states are
to be calculated for a larger system.
9.18.7 Natural Transition Orbitals
Results of TD-DFT or CIS calculations can be tedious to interprete as many individual MO pairs may
contribute to a given excited state. In order to facilitate the analysis while keeping the familiar picture of an
excited state originating from essentially an electron being promoted from a donor orbital to an acceptor
orbital, the device of natural transition orbitals can be used.
The procedure is quite straightforward. For example, consider the following job on the pyridine molecule:
! PBE D3ZERO def2-SVPD def2/J tightscf
%tddft nroots 5
DoNTO true # flag to turn on generation of natural transition orbitals
NTOStates 1,2,3 # States to consider for NTO analysis;
#if empty all will be done
NTOThresh 1e-4 # threshold for printing occupation numbers
end
* xyz 0 1
N 0.000000 0.000000 1.401146
C 0.000000 1.146916 0.702130
C 0.000000 -1.146916 0.702130
C -0.000000 1.205574 -0.702848
C -0.000000 -1.205574 -0.702848
C 0.000000 -0.000000 -1.421344
H -0.000000 2.079900 1.297897
H -0.000000 -2.079900 1.297897
H -0.000000 2.179600 -1.219940
H -0.000000 -2.179600 -1.219940
H 0.000000 0.000000 -2.525017
*
which results in:

------------------------------------------
NATURAL TRANSITION ORBITALS FOR STATE 1
------------------------------------------
Making the (pseudo)densities ... done

Solving eigenvalue problem for the occupied space ... done
Solving eigenvalue problem for the virtual space ... done
Natural Transition Orbitals were saved in TD-DFT-Example-6.s1.nto
Threshold for printing occupation numbers 0.000100
E= 0.158492 au 4.313 eV 34785.0 cm**-1

20a -> 21a : n= 0.99825296
19a -> 22a : n= 0.00067172
18a -> 23a : n= 0.00051394
17a -> 24a : n= 0.00030846
------------------------------------------
------------------------------------------
Making the (pseudo)densities ... done

Solving eigenvalue problem for the occupied space ... done
Solving eigenvalue problem for the virtual space ... done
Natural Transition Orbitals were saved in TD-DFT-Example-6.s3.nto
Threshold for printing occupation numbers 0.000100
E= 0.197103 au 5.363 eV 43259.2 cm**-1

20a -> 21a : n= 0.64493520
19a -> 22a : n= 0.34962356
18a -> 23a : n= 0.00166855
17a -> 24a : n= 0.00112178
16a -> 25a : n= 0.00073279
15a -> 26a : n= 0.00062556
14a -> 27a : n= 0.00045127
13a -> 28a : n= 0.00023135
12a -> 29a : n= 0.00019911
11a -> 30a : n= 0.00017459
10a -> 31a : n= 0.00011544
-----------------------------------------------------------------------------
ABSORPTION SPECTRUM VIA TRANSITION ELECTRIC DIPOLE MOMENTS
-----------------------------------------------------------------------------
State Energy Wavelength fosc T2 TX TY TZ
(cm-1) (nm) (au**2) (au) (au) (au)
-----------------------------------------------------------------------------
1 34785.0 287.5 0.004079502 0.03861 -0.19649 0.00000 -0.00000
2 35118.0 284.8 0.000000000 0.00000 0.00000 0.00000 0.00000
3 43259.2 231.2 0.024852699 0.18913 -0.00000 0.43490 -0.00000
4 49592.3 201.6 0.000013122 0.00009 0.00000 0.00000 -0.00933
5 54405.8 183.8 0.027126662 0.16414 -0.00000 -0.40515 -0.00000
We see that there is a weakly allowed transition (S1) that is essentially totally composed of a single NTO pair
(20a21a : n= 0.99825296), while the third excited state (S3) is strongly allowed and requires two NTO
pairs for its description (20a21a : n= 0.64493520 and 19a22a : n= 0.34962356).
These orbitals are shown below. It is evident that the S1 state donor orbital (NTO10) is a nitrogen lone
pair and the acceptor orbital is a orbital of the ring. For the S3 state the two NTO donor orbitals are
comprised of a nearly degenerate set of orbitals (they would be degenerate in the parent benzene) and
the acceptor orbitals are a pair of nearly degenerate orbitals. It is evident from this example that by
looking at the NTOs one can obtain a nicely pictorial view of the transition process, even if many orbital
pairs contribute to a given excited state in the canonical basis.
Figure 9.21: Natural transition orbitals for the pyridine molecule in the S1 and S3 states.
Similar analysis can be performed in the case of ROCIS and DFT/ROCIS calculations as it will be described
in section 9.19.3.
9.18.8 Computational Aspects
9.18.8.1 RI Approximation (AO-Basis)
If the SCF calculation used the RI approximation it will also be used in the TD-DFT calculation. The RI
approximation saves a large amount of time while giving close to identical results (the errors will usually be
<0.1 eV) and is generally recommended. If the functional is a hybrid functional the RI-approximation will
only be applied to the Coulomb term while the exchange will be treated as before. In the SCF you can use
this feature with the keyword (! RIJONX). It will then also be used in the TD-DFT calculation. Again, the
RIJCOSX approximation can be used in TD-DFT and CIS calculations and leads to very large speedups at
virtually no loss in accuracy.
9.18.8.2 RI Approximation (MO-Basis)
As an alternative to the direct AO-basis computation ORCA allows to use RI-integrals transformed to the
MO basis to generate the CI matrix. This algorithm is more disk-intensive. However, for medium sized
molecules we have observed speedups on the order of 15 or more with this method. It is particularly benefitial
together with hybrid functionals.
In order to use this method you have to specify mode riints in the %tddft block and you also have to assign
an auxiliary basis set (for example def2-TZVP/C). There is a second algorithm of this kind that is labelled
mode riints disk
Note that the auxiliary basis set has to be valid for correlation treatments in case that you have a hy-
brid functional. Thus the basis sets developed for RI-MP2 are suitable (def2-SVP/C, def2-TZVP/C and
def2-TZVPP/C). If you have a non-hybrid functional the normal RI-J auxiliary basis sets are fine.
An example that uses the B3LYP functional is given below:
! RKS B3LYP/G SV(P) SV/C TightSCF
%tddft
mode riints # or riints disk (often faster but requires more disk space)
nroots 8
end
* int 0 1
C 0 0 0 0.00 0.0 0.0
O 1 0 0 1.20 0.0 0.0
H 1 2 0 1.08 120.0 0.0
H 1 2 3 1.08 120.0 180.0
*
NOTE:
Do not forget to assign a suitable auxiliary basis set! If Hartree-Fock exchange is present (HF or
hybrid-DFT) these are the auxiliary bases optimized for correlation while for non-hybrid functionals
the standard RI-J bases are suitable.
The standard auxiliary basis sets may not be suitable if you have diffuse functions present and want to
study Rydberg states. You have to augment the axuliary basis with diffuse functions yourself in this
case.
Be prepared that the transformed integrals take up significant amounts of disk space.
9.18.8.3 Integral Handling
If the SCF calculation is carried out in an integral direct fashion this will also be done in the CIS/TD-DFT
calculation. Thus, no bottlenecks arising from large integral transformations or large disk space requirement
arise in the calculations. An exception is the MO based RI approximations described in the previous section.
9.18.8.4 Valence versus Rydberg States
For valence excited states the usual orbital basis sets are reasonable. Thus, with polarized double-zeta basis
sets sensible results are obtained. Especially DFT calculations have the nice feature of not being overly basis
set dependent.
If Rydberg states are desired you should make sure that diffuse functions are present in your basis set. These
can be added to any normal basis set. For example, the following example provides a rather high quality
basis for excited state calculations that is based on the Ahlrichs basis set:
%basis basis vtz # standard triple-zeta

pol Ahlrichs_2df # large (2d1f) polarization
diff _p # plus 1 s,p diffuse set
# augment the carbon basis set by even more
# diffuse functions
addgto 6
s 1
1 0.01 1.0
p 1
1 0.01 1.0
d 1
1 0.07 1.0
end
end
Smaller basis sets may also be sufficient.
TIP
If you want to augment a given basis set it is sensible to run a preliminary SCF calculation and use
%output print[p basis] 2 end. This will provide you with a detailed listing of basis functions and
their exponents. You can then add additional s, p and perhaps d-functions with the AddGTO command
as in the example above. It is sensible to decrease the exponent of the diffuse functions by roughly a
factor of 3 from the smallest exponent in the original basis.
9.18.8.5 Restrictions for Range-Separated Density Functionals
Several restrictions apply for range-separated density functionals. They are currently only implemented to
work with the AO-based algorithm within the RIJONX, RIJCOSX, and NORI integral schemes. Moreover,
the calculation of triplet states has been disabled. Additionally, the asymptotic correction (section 9.18.8.6)
has been disabled. However, the nuclear gradient for the excited states is now available.
9.18.8.6 Asymptotatically Corrected Density Functionals
In studying Rydberg states one general shortcoming of the present day density functionals becomes particularly
prominent. This is the too fast decay of the DFT potential in the long range limit. This causes the DFT
orbital energies being too positive by several eV. Likewise the excitation energies to Rydberg states are
underestimated by several eV. This can be fixed by using density functionals with improved long range
behavior. One such functional is the LB94 functional. However, the LB94 has the disadvantage of being only
defined through the potential and that this potential is not accurate in the bonding region. Therefore it is
reasonable to only correct an existing functional (like the BP functional) in the long range with the LB94
functional. In ORCA the so-called gradient regular asymptotic correction (GRAC) is implemented and
can be applied with any bulk density functional in the bonding region. The energetics and occupied orbital
energies are only affected to a minor extent while the virtual orbital energies are significantly changed and
the excitation energies to Rydberg states are improved.
The following example shows how to use the asymptotic correction. The run almost exactly reproduces the
uning et al. [380] who also developed the method implemented in ORCA.
results of Gr
#
# Gradient-regulated asymptotic correction calculation
# on the excited states of the CO molecule.
#
# Reference: Gruning et al. (2001) JCP, vol. 114, p. 652
#
! RKS BP NoRI Grid5 NoFinalGrid TightSCF
%method xckernel lda
ldaopt c_vwn5
xckernellda c_vwn5
# *** turn on the asympt. correction
gracLB true
# *** the input ionization potential
# *** in eV !!!
ip 14.01
end
%basis basis "tzvpp++"
addgto 6
s 1
1 0.01 1.0
p 1
1 0.01 1.0
d 1
1 0.07 1.0
end
addgto 8
s 1
1 0.02 1.0
p 1
1 0.02 1.0
d 1
1 0.1 1.0
end
end
%tddft nroots 10 # no of roots to determine
tda false # Tamm-Dancoff approx
etol -1 # energy tolerance
rtol 1e-5 # residual tolerance
triplets true # generate triplets ?
end
* xyz 0 1
C 0 0 0.00
O 0 0 1.130
*
From the example two additional inputs are necessary compared to a standard DFT calculation. The first is
gracLB true in the method block to turn on the asymptotic correction. The second is ip 14.01 also in
the method block. The second input is the ionization potential of the molecule in eV and is required for the
correction to be meaningful. If there is no experimental value available you have to calculate the IP first by a
SCF calculation before you can run the TD-DFT!
CAUTION:
The corrected density functional potentials are not functional derivatives of well defined energy
functionals. Therefore you should not rely on the total energies delivered by these functionals because
these are not variational energies!
9.18.8.7 Potential Energy Surface Scans
ORCA allows the combination the scan feature with CIS or TD-DFT. This can be used to map out the
excited state potential energy surfaces as a function of one- two- or three parameters. The output of the
trajectory run automatically contains the excited state energies in addition to the ground state energy. For
example consider the following simple job.
! def2-TZVPD
%method scanguess pmodel # this assignment forces a PModel guess at each step
# which is often better if diffuse functions are present
end
%cis NRoots 7
end
%paras rCO = 0.85,1.45,21;
end
* xyz 0 1
O 0 0 0
C 0 0 {rCO}
*
The output file from this job contains the total energies (i.e. the ground state energy plus the excitation
energy) for each excited state as a function of C-O bondlength as shown below. Howerver, the assignment of
the individual states will change with geometry due to curve crossings. Thus, the state-to-state correlation
must be worked out by hand. These calculations are nevertheless very helpful in obtaining at least a rough
idea about excited state energy surfaces.
Figure 9.22: Result of a potential energy surface scan for the excited states of the CO molecule
using the orca cis module.
9.18.8.8 Potential Energy Surface Scans along Normal Coordinates
The ground and excited state potential energy surfaces can also be mapped as a function of normal coordinates.
The normal mode trajectory run is invoked by the keyword !MTR. In addition several parameters have to be
specified in the block %mtr. The following example illustrates the use:
First you run a frequency job:
#
! BP86 def2-SV(P) def2/J TightSCF AnFreq
* xyz 0 1
C 0.000001 -0.000000 -0.671602
C 0.000000 0.000000 0.671602
H -0.000000 -0.940772 -1.252732
H -0.000000 -0.940772 1.252732
H -0.000000 0.940772 -1.252732
H -0.000000 0.940772 1.252732
*
and then:
! BP86 def2-SV(P) def2/J TightSCF MTR
%tddft
NRoots 3
triplets false
end
%mtr
HessName "ethene.hess"
modetype normal
MList 9,13
RSteps 4,5
LSteps 4,5
ddnc 1.0, 0.5
end
* xyz 0 1
C 0.000001 -0.000000 -0.671602
C 0.000000 0.000000 0.671602
H -0.000000 -0.940772 -1.252732
H -0.000000 -0.940772 1.252732
H -0.000000 0.940772 -1.252732
H -0.000000 0.940772 1.252732
*
The HessName parameter specifies the name of the file which contains nuclear Hessian matrix calculated
in the frequency run. The Hessian matrix is used to construct normal mode trajectories. The keyword
MList provides the list of the normal modes to be scanned. The parameters RSteps and LSteps specify the
number of steps in positive and negative direction along each mode in the list. In general, for a given set of
parameters
mlist m1,m2,...mn
rsteps rm1,rm2,...rmn
lsteps lm1,lm2,...lmn
the total number of the displaced geometries for which single point calculations will be performed is equal to
Q
(rmi + lmi + 1). Thus, in the present case this number is equal to (4 + 4 + 1) (5 + 5 + 1) = 99.
mi
The ddnc parameter specifies increments q for respective normal modes in the list in terms of dimensionless
normal coordinates (DNCs). The trajectories are constructed so that corresponding normal coordinates are
varied in the range from l q to r q . The measure of normal mode displacements in terms DNCs is
appropriate choice since in spectroscopical applications the potential energy function U is usually expressed
in terms of the DNCs. In particular, in the harmonic approximation U (q ) has a very simple form around
equilibrium geometry:
3N 6
X } 2
U = U0 + q (9.249)

2
where is the vibrational frequency of the -th mode.
Dimensionless normal coordinate q can be related to the vector of atomic Cartesian displacements X as
follows:
3N
21 X

p
q = Lk Xk Mk (9.250)
}
k=1
where {Lk } is the orthogonal matrix obtained upon numerical diagonalization of the mass-weighted Hessian
matrix, and M is the vector of atomic masses. Accordingly, the atomic Cartesian displacements corresponding
to a given dimensionless normal coordinate q are given by:
12
} 21
Xk = Lk q (Mk ) (9.251)

Alternatively, it is possible to specify in the input the Cartesian increment for each normal mode. In such a
case, instead of the ddnc parameter one should use the dxyz keyword followed by the values of Cartesian
displacements, for example:
%mtr
modetype normal
MList 9,13
RSteps 4,5
LSteps 4,5
dxyz 0.01, 0.02 # increments in the Cartesian basis
# are given in angstrom units

end
For a given Cartesian increment dX, along the th normal mode the atomic displacements are calculated
as follows:
dX, 1
Xk = Lk (Mk ) 2 (9.252)
kT k
12
The vector T in the Cartesian basis has components Ti = Lk (Mk ) and length (norm) kTk k.
The increment length can also be selected on the basis of an estimate for the expected change in the total
energy E due to the displacement according to eq.9.123. The value of E can be specified via the EnStep
parameter:
%mtr
modetype normal
MList 9,13
RSteps 4,5
LSteps 4,5
EnStep 0.001, 0.001 # the values are given in Eh
end
All quantum chemical methods have to tolerate a certain amount of numerical noise that results from finite
convergence tolerances or other cutoffs that are introduced into the theoretical procedures. Hence, it is
reasonable to choose E such that it is above the characteristic numerical noise level for the given method of
calculation.
At the beginning of the program run the following trajectory files which can be visualized in gOpenMol will
be created:
BaseName.m9.xyz and BaseName.m13.xyz contain trajectories along normal modes 9 and 13, respec-
tively.
BaseName.m13s1.m9.xyz - BaseName.m13s5.m9.xyz contain trajectories along normal mode 9 for

different fixed displacements along mode 13, so that the file BaseName.m13sn.m9.xyz corresponds to
the n-th step in the positive direction along mode 13.
BaseName.m13s-1.m9.xyz - BaseName.m13s-5.m9.xyz contain trajectories along normal mode 9 for

different fixed displacements along mode 13, so that the file BaseName.m13s-n.m9.xyz corresponds to
the n-th step in the negative direction along mode 13.
BaseName.m9s1.m13.xyz - BaseName.m9s4.m13.xyz contain trajectories along normal mode 13 for

different fixed displacements along mode 9, so that the file BaseName.m9sn.m13.xyz corresponds to
the n-th step in the positive direction along mode 9.
BaseName.m9s-1.m13.xyz - BaseName.m9s-4.m13.xyz contain trajectories along normal mode 13

for different fixed displacements along mode 9, so that the file BaseName.m9s-n.m13.xyz corresponds
to the n-th step in the negative direction along mode 9.
The results of energy single point calculations along the trajectories will be collected in files BaseName.mtr.escf.S.dat
(for the SCF total energies) and files BaseName.mtr.ecis.S.dat (for the CIS/TDDFT total energies), where
S in the suffix of *.S.dat filenames provides specification of the corresponding trajectory in the same
way as it was done for the case of trajectory files *.xyz (e.g. S=m9s-1.m13). Likewise, the cal-
culated total energies along the trajectories will be collected in files BaseName.mtr.emp2.S.dat in the
case of MP2 calculations, BaseName.mtr.emdci.S.dat (MDCI), BaseName.mtr.ecasscf.S.dat (CASSCF),
BaseName.mtr.emrci.S.dat (MRCI).
Note, that in principle normal coordinate trajectories can be performed for an arbitrary number normal modes.
This implies that in general trajectories will contain geometries which involve simultataneous displacement
along several (>2) modes. However, trajectory files *.xyz and corresponding *.dat files will be generated
only for the structures which are simultaneously displaced along not more than 2 normal coordinates.
Figure 9.23: Result of a potential energy surface scan along C-C stretching normal coordinate
(mode 13 in the present example) for the excited states of the ethene molecule using
the orca cis module.
9.18.8.9 Normal Mode Scan Calculations Between Different Structures
This type of job allows to map PES between two different structures as a function of normal coordinates.
The H2 O molecule represent a trivial case which has formally 2 equivalent equilibrium structures which differ
by angle H1 OH2 ( 103.5 and 256.5 , respectively, as follows from the BP86/SV(P) calculations). In
such a case the input for the nomal mode trajectory run would require the calculation of geometry difference
between both structures in terms of the dimensionless normal coordinates. This can be done in orca vib run
as follows :
> orca_vib water.hess ddnc geom.xyz
The second parameter ddnc in the command line invokes the calculation of geometry difference in terms of
the DNCs. Both structures are specified in the file geom.xyz which has a strict format:
2 3
0
0.000000 0.000000 0.000000
0.000000 0.607566 0.770693
0.000000 0.607566 -0.770693
1
0.000000 0.000000 0.000000
0.000000 -0.607566 0.770693
0.000000 -0.607566 -0.770693
The first line of the input specifies the number of the structures and total number of atoms (2 and 3,
respectively). Specification of each structure in sequence starts with a new line containing the number of
the structure. The number 0 in the second line is used to denote the reference structure. Note that atomic
coordinates should be given in units of
A and in the same order as in the ORCA input for the frequency run
from which the file water.hess was calculated.
At the end of the orca vib run the file geom.ddnc is generated. It contains the geometry difference in terms
of the dimensionless normal coordinates between the structures with nonzero numbers and the reference one
in geom.xyz :
1
1 9
0 0.000000
1 0.000000
2 0.000000
3 0.000000
4 0.000000
5 0.000000
6 9.091932
7 -9.723073
8 0.000000
9.19 Excited States via ROCIS and DFT/ROCIS 521
The output file indicates that the structural difference occurs along 2 normal coordinates: 6 (bending mode)
and 7 (totally symmetric OH stretching mode). On the basis of the calculated displacement pattern the
following input for the normal mode trajectory run between two structures can be designed:
! RKS BP86 SV(P) def2/J RI TightScf MTR
%mtr
HessName "water.hess"
modetype normal
mlist 6,7
rsteps 10,0
lsteps 0, 10
ddnc 0.9091932, 0.9723073
end
* xyz 0 1
O 0.000000 0.000000 0.000000
H 0.000000 0.607566 0.770693
H 0.000000 0.607566 -0.770693
*
Here the parameters RSteps, LSteps and ddnc are chosen in such a way that in the scan along modes 6 and 7
the corresponding dimensionless normal coordinates will be varied in the range 0 9.091932 and -9.723073
0, respectively, in accordance with the projection pattern indicated in the file geom.ddnc. Note that normal
modes are only defined up to an arbitrary choice of sign. Consequently, the absolute sign of the dimensionless
displacements is ambiguous and in principle can vary in different orca vib runs. It is important that the
normal mode scan between different structures exemplified above is performed using the same sign of normal
modes as in the calculation of normal mode displacements. This condition is fulfilled if the same normal
modes are used in orca vib run and trajectory calculation. Thus, since in orca vib calculation normal
modes are stored in .hess file it is necessary to use the same Hessian file in the trajectory calculation.
9.19 Excited States via ROCIS and DFT/ROCIS
The ORCA program package includes the orca rocis module to perform configuration interaction with
single excitations (CIS) calculations using a restricted open-shell Hartee-Fock (ROHF) reference function.
It produces excitation energies, absorption energies and CD intensities. It was designed with the aim to
reproduce and - even more importantly - reliably predict transition metal L-edges as observed in X-ray
absorption spectroscopy (XAS).
9.19.1 General Use
In the present implementation the orca rocis module is only able to perform CIS calculations on top of a
high-spin ROHF reference function. All spins of the unpaired electrons have to be coupled ferrmoagnetically
to give a total spin of S = 12 N , where N is the number of unpaired electrons. Other ROHF functions such as
Zerners configuration averaged or spin averaged ROHF cannot be used as reference. The input for a high
spin ROHF calculation is done in the %scf block.
%scf HFTyp ROHF #Flag for ROHF

ROHF_Case HighSpin #selects the high-spin case
ROHF_NEl[1] = 4 #the number of unpaired electrons
end
In our experience ROHF calculations suffer a lot from convergence problems. UHF calculations generally
exhibit better convergence properties. In most cases the quasi-restricted orbitals (qros) of a UHF calculation
resemble the ROHF orbitals. Thus the program features the ability to start a ROCIS calculation on top of
a UHF calculation. It will automatically create the qros and build the reference determinant with them.
If one wants to avoid the (small) errors that are introduced by this procedure, one may take the qros of a
UHF calculation as starting orbitals for a subsequent ROHF calculation. Furthermore it is possible to invoke
the orca rocis module for closed-shell molecules. The program will then perform a CI calculation with the
provided RHF reference function. In this case it will yield the same result as the orca cis program.
A number of basic variables in the %rocis block control the settings of the Davidson procedure that is used
to solve the CI problem:
%rocis NRoots 6 # number of desired roots

ETol 1e-6 # energy convergence tolerance
RTol 1e-6 # residual vector convergence tolerance
MaxIter 35 # maxmimum number of iterations
NGuessMat 512 # dimension of the guess matrix: 512x512
end
The dimension of the iterative subspace is given by MaxDim cdot NRoots. The lowest possible choice for
MaxDim is a value of 2. In general, by choosing MaxDim 5-10 times NRoots you will achieve a more
favorable convergence by the cost of an increased disk space requirement. Increasing the NGuessMat variable
will improve the convergence of the iterative CI procedure. The amount of output produced during the
calculation is controlled via the PrintLevel variable
%rocis NRoots 3
PrintLevel 3
end
Note, that this does not influence which spectra are calculated or printed. The absorption spectrum calculated
on the basis of the pure dipole approximation for your calculation is always printed. In addition, it is possible
to allow for electric quadrupole and magnetic dipole contributions to the absorption spectrum as well as to
calculate the CD spectrum:
%rocis NRoots 6
DoQuad true #invokes calculation of electric
#quadrupole and magnetic dipole
#magnetic quadrupole and electric octupole contributions
DoCD true #invokes the calculation of the CD
#spectrum
end
The printed spectra look like this:
-----------------------------------------------------------------------------
-----------------------------------------------------------------------------
(cm-1) (nm) (au**2) (au) (au) (au)
-----------------------------------------------------------------------------
1 2635.0 3795.1 0.000000001 0.00000 0.00001 -0.00001 0.00029
2 4365.5 2290.7 0.000011416 0.00086 0.01200 -0.00864 0.02534
3 4368.2 2289.3 0.000011174 0.00084 -0.02006 0.01442 0.01523
4 5977.9 1672.8 0.000093897 0.00517 -0.04164 -0.05863 0.00000
5 65245.3 153.3 0.027669631 0.13961 -0.20555 -0.31203 -0.00023
-----------------------------------------------------------------------------
ABSORPTION SPECTRUM VIA TRANSITION VELOCITY DIPOLE MOMENTS
-----------------------------------------------------------------------------
State Energy Wavelength fosc P2 PX PY PZ
(cm-1) (nm) (au**2) (au) (au) (au)
-----------------------------------------------------------------------------
1 2635.0 3795.1 0.000000085 0.00000 -0.00000 0.00000 -0.00004
2 4365.5 2290.7 0.001777771 0.00005 -0.00315 0.00223 -0.00618
3 4368.2 2289.3 0.001850956 0.00006 0.00526 -0.00372 -0.00371
4 5977.9 1672.8 0.003237195 0.00013 0.00667 0.00937 0.00000
5 65245.3 153.3 0.057301314 0.02555 0.08779 0.13358 0.00010
-------------------------------------------------------------------
CD SPECTRUM
-------------------------------------------------------------------
State Energy Wavelength R MX MY MZ
(cm-1) (nm) (1e40*sgs) (au) (au) (au)
-------------------------------------------------------------------
1 2635.0 3795.1 0.00007 -0.00511 -0.01539 0.00021
2 4365.5 2290.7 10.02484 0.57434 -0.40490 0.42899
3 4368.2 2289.3 -10.03730 0.34432 -0.24269 -0.71470
4 5977.9 1672.8 0.01537 -0.00033 -0.00032 -0.00286
5 65245.3 153.3 -0.00865 0.00004 0.00003 -0.00005
-----------------------------------------------------------------------------------------------------
-----------------------------------------------------------------------------------------------------
State Energy Wavelength D2 m2 Q2 D2+m2+Q2 D2/TOT m2/TOT Q2/TOT
(cm-1) (nm) (*1e6) (*1e6)
-----------------------------------------------------------------------------------------------------
1 2635.0 3795.1 0.00000 0.00011 0.00000 0.00000000080469 0.86010 0.13938 0.00052
2 4365.5 2290.7 0.00001 0.47866 0.00000 0.00001189497194 0.95976 0.04024 0.00000

3 4368.2 2289.3 0.00001 0.48629 0.00000 0.00001166062671 0.95830 0.04170 0.00000
4 5977.9 1672.8 0.00009 0.00001 0.00001 0.00009389664707 1.00000 0.00000 0.00000
5 65245.3 153.3 0.02767 0.00000 0.06183 0.02766969236508 1.00000 0.00000 0.00000
-----------------------------------------------------------------------------------------------------
COMBINED ELECTRIC DIPOLE + MAGNETIC DIPOLE + ELECTRIC QUADRUPOLE SPECTRUM (origin adjusted)
-----------------------------------------------------------------------------------------------------
State Energy Wavelength D2 m2 Q2 D2+m2+Q2 D2/TOT M2/TOT Q2/TOT
(cm-1) (nm) (*1e6) (*1e6)
-----------------------------------------------------------------------------------------------------
1 2635.0 3795.1 0.00000 0.00000 0.00000 0.00000000069409 0.99716 0.00016 0.00268
2 4365.5 2290.7 0.00001 0.38277 0.00039 0.00001179947536 0.96753 0.03244 0.00003
3 4368.2 2289.3 0.00001 0.36798 0.00045 0.00001154275975 0.96808 0.03188 0.00004
4 5977.9 1672.8 0.00009 0.00000 0.00001 0.00009389663928 1.00000 0.00000 0.00000
5 65245.3 153.3 0.02767 0.00003 0.06176 0.02766969232228 1.00000 0.00000 0.00000
Furthermore like in TD-DFT one may obtain nearly origin independent intensities by evaluating in addition
to the electric dipole (D), electric quadrupole (Q) and magnetic dipole (m) intensities the corresponding
electric dipole - magnetic quadrupole (DM) and the electric dipole - electric octupole (DO) intensities. Hence
in addition to the above spectra one obtains
----------------------------------------------------------------------------------------------
(Origin Independent, Length Representation)
----------------------------------------------------------------------------------------------
State Energy Wavelength D2 m2 Q2 DM DO D2+m2+Q2+DM+DO ...
(cm-1) (nm) (*1e6) (*1e6) (*1e6) (*1e6) (*1e6)
----------------------------------------------------------------------------------------------
....
These spectra are plotted by calling
orca_mapspc MyOutput.out ABS/ABSV/CD/ABSQ/ABSOI -eV -x0(start) -x1(stop) -w(width) -n(points)
If calculations on large molecules are conducted, the integral transformation will be the most time-consuming
part. Therefore it is strongly recommended to use the resolution of the identity (RI) approximation in those
cases. It effectively reduces the computational costs of the transformation step by only introducing minor
errors to the calculation. It has to be kept in mind that in order to keep the introduced errors small, one has
to provide a reasonable auxiliary basis sets along with your normal basis set input.
! TZVP TZV/C TightSCF SlowConv
%SCF HFTyp ROHF

ROHF_Case HighSpin
ROHF_Nel[1] = 1
End
%ROCIS NROOTS 5
DoRI true # invokes the RI approximation
DoQuad true
end
* xyz 0 2
N 0 0 0
O 0 0 1.15
*
The orca rocis module provides two ways of choosing the orbital excitation space: by orbital energy or
orbital number. In the former case an energy window has to be specified and the program will then take all
orbitals, whose orbital energies lie within this window, into account. Note, that one actually has to define
two orbital windows: One for the donor and the second for the acceptor orbital. The input of the windows
is done as an array: The first two numbers define the donor space while the last two numbers define the
acceptor space.
%rocis NRoots 3
EWin = -5,5,-5,5
end
The default is to keep core orbitals and very high lying virtual orbitals out of their respective orbital excitation
spaces. Since these orbitals span a space that is usually not reachable with regular UV/Vis spectroscopy, this
is a reasonable approximation. One has to keep in mind that an orbital energy window makes only sense if
the orbitals used in the calculation have a well-defined orbital energy. As a consequence one cannot use an
orbital energy window for a calculation with localized orbitals. The second way to specify the excitation
space is by orbital numbering.
%rocis NRoots 3
OrbWin = 1,13,9,22
end
In restricted calculations only one set of spatial orbitals is created. Hence it is not necessary to provide
orbital windows for and electrons separately. Of course, only doubly or singly occupied orbitals can
act as donor orbitals and only singly and nonoccupied orbitals can act as acceptor orbitals. The program
recognizes nonoccupied orbitals in the donor space and doubly occupied orbitals in the acceptor space and
removes both.
The many-electron expansion space of a ROCIS calculation in ORCA is divided into five classes. Using
q a
second quantized replacement operators Epq = a q a
p + a p they take the form [381].
|si i = Eis |0i

|as i = Esa |0i
|ai i = 12 Eia |0i (9.253)
|as
ti i = Eta Eis |0i
as
|ti i = 1 (E a 2Esa Eis ) |0i
6 i
The orbital label i denotes a doubly occupied orbital, s and t refer to singly occupied orbitals and orbital
label a corresponds to a virtual orbital. The form of the excitation classes ensures that all excited states are
eigenfunctions of the S2 -operator and have the same total spin S as the electronic ground state. Each of the
five excitation classes can be switched on or off manually.
%rocis NRoots 3
Do_is true #Include DOMO->SOMO excitations
Do_sa true #Include SOMO->Virtual excitations
Do_ia true #Include DOMO->Virtual excitations
Do_ista true #Include DOMO->SOMO coupled to
#SOMO->Virtual excitations with #s not equal t
Do_isa true #Include DOMO->SOMO coupled to
#SOMO->Virtual excitations with #s = t
#---------------------------------
#by default all switches for the
#excitation classes are set to
#true
#---------------------------------
end
Formally, the |as at

ti i and |ti i excitation classes can be regarded as double excitations. When the program
finishes the ROCIS calculation it gives the excitation energy together with the composition for each root.
According to the number of labels of the respective functions |i, contributions from excited configuration
state functions belonging to the different excitation classes are given by two, three or four numbers.
STATE 5 Exc. Energy: 297.279mEh 8.089eV 65245.3cm**-1

47->50 : 0.2196
47->51 : 0.0138
37->50 : 0.1165
41->50 : 0.0960
38->46 ; 47->50 : 0.0103
37->46 ->50 : 0.0150
37->47 ->50 : 0.0938
37->48 ->50 : 0.0179
37->49 ->50 : 0.0179
41->46 ->50 : 0.0174
41->47 ->50 : 0.0585
41->48 ->50 : 0.0213
41->49 ->50 : 0.0211
Furthermore the orca rocis module is able to calculate the effect of spin-orbit coupling (SOC) on the
calculated ground and excited states. It introduces SOC in the framework of quasi-degenerate perturbation

theory (QDPT). The SOC Hamiltonian is diagonalized in the basis of the calculated ROCIS states SM I ,
where I is the root label and S and M are the spin and magnetic spin quantum numbers, respectively [162],
[381].
%rocis NRoots 3
OrbWin = 1, 3 ,9 ,22
SOC true #invokes the calculation of #SOC effects
SOCTEMP 10 #temperature for SOC #corrected spectra in Kelvin
end
After the SOC calculation the program will produce additional spectra for the SOC corrected results. The
spectra contain transitions from the 2S + 1 lowest lying states into all excited states, where S is the spin
quantum number of the electronic ground state. These 2S + 1 lowest states may be split up in the order of
1-100 cm1 . Due to the small magnitude of the splitting, all of the 2S+1 states can be significantly populated
even at low temperatures. Experimentally, the intensity of a given transition is dependent on the population
of the corresponding initial state. With the SOCTemp keyword the population of the theoretically calculated
states can be manipulated by the varying the fictive temperature of the system. It has to be mentioned that
the electric quadrupole transitions between spin-orbit coupled states are not well defined and are likely to
give unreasonable results. Hence it is recommended to use the DoQuad keyword only for calculations that do
not include SOC.
-------------------------------------------------------------------------------
SPIN ORBIT CORRECTED ABSORPTION SPECTRUM VIA TRANSITION ELECTRIC DIPOLE MOMENTS
-------------------------------------------------------------------------------
(cm-1) (nm) (au**2) (au) (au) (au)
-------------------------------------------------------------------------------
0 1 5.6 0.0 0.000000000 0.00000 0.00003 0.00002 0.00000
0 2 6.2 0.0 0.000000000 0.00000 0.00000 0.00000 0.00005
0 3 23.7 422287.3 0.000000000 0.00000 0.00000 0.00000 0.00000
0 4 23.7 421562.8 0.000000000 0.00000 0.00018 0.00025 0.00000
0 5 2621.7 3814.3 0.000000000 0.00000 0.00000 0.00001 0.00005
0 6 2622.0 3813.9 0.000000000 0.00000 0.00003 0.00012 0.00000
0 7 2634.7 3795.5 0.000000095 0.00002 0.00388 0.00273 0.00049
0 8 2634.9 3795.2 0.000000103 0.00002 0.00039 0.00027 0.00495
0 9 2639.5 3788.6 0.000000001 0.00000 0.00001 0.00001 0.00036
0 10 4223.6 2367.6 0.000000103 0.00002 0.00043 0.00029 0.00390
0 11 4223.9 2367.5 0.000000120 0.00002 0.00348 0.00236 0.00046
0 12 4296.3 2327.6 0.000000696 0.00010 0.00562 0.00842 0.00000
0 13 4357.6 2294.8 0.000000002 0.00000 0.00001 0.00001 0.00049
0 14 4418.1 2263.4 0.000005778 0.00083 0.00653 0.00468 0.02762
0 15 4422.1 2261.4 0.000005517 0.00079 0.02184 0.01559 0.00832
0 16 4488.2 2228.0 0.000000001 0.00000 0.00004 0.00006 0.00038
0 17 4524.2 2210.3 0.000000001 0.00000 0.00030 0.00018 0.00000
0 18 4597.2 2175.2 0.000000027 0.00000 0.00023 0.00016 0.00191
0 19 4597.4 2175.2 0.000000051 0.00001 0.00213 0.00153 0.00023
0 20 6043.6 1654.6 0.000047989 0.00502 0.04104 0.05779 0.00000
0 21 6049.5 1653.0 0.000000014 0.00000 0.00109 0.00057 0.00001
0 22 6051.3 1652.5 0.000000021 0.00000 0.00001 0.00004 0.00150

0 23 6069.7 1647.5 0.000000000 0.00000 0.00005 0.00007 0.00000
0 24 6069.9 1647.5 0.000000028 0.00000 0.00098 0.00138 0.00000
0 25 65281.7 153.2 0.014223474 0.13787 0.20423 0.31010 0.00023
0 26 65281.7 153.2 0.000000035 0.00000 0.00032 0.00048 0.00011
0 27 65281.7 153.2 0.000009000 0.00009 0.00522 0.00774 0.00001
0 28 65281.7 153.2 0.000007207 0.00007 0.00460 0.00698 0.00000
0 29 65281.7 153.2 0.000047448 0.00046 0.01179 0.01791 0.00001
1 2 0.6 0.0 0.000000000 0.00000 0.00001 0.00001 0.00000
1 3 18.1 553477.5 0.000000000 0.00000 0.00000 0.00000 0.00009
1 4 18.1 552233.6 0.000000000 0.00000 0.00006 0.00004 0.00000
1 5 2616.1 3822.5 0.000000063 0.00001 0.00006 0.00003 0.00261
1 6 2616.4 3822.1 0.000000060 0.00001 0.00211 0.00144 0.00006
1 7 2629.1 3803.6 0.000000143 0.00002 0.00225 0.00321 0.00003
1 8 2629.3 3803.3 0.000000002 0.00000 0.00015 0.00025 0.00040
1 9 2633.9 3796.7 0.000000271 0.00003 0.00011 0.00008 0.00538
1 10 4218.0 2370.8 0.000000005 0.00000 0.00031 0.00046 0.00019
If the PrintLevel value is set to 3 or higher, the program will print out the composition of the SOC corrected

states in the basis of states SM
I .
Eigenvectors of SOC calculation:

the threshold for printing is: 0.010000
weight : Root Spin Ms
State 0: 0.00 cm**-1 0.00000 eV
0.378045 : 0 2 2
0.235825 : 0 2 0
0.378045 : 0 2 -2
State 1: 5.61 cm**-1 0.00070 eV

0.496236 : 0 2 2
0.496236 : 0 2 -2
State 2: 6.20 cm**-1 0.00077 eV

0.496291 : 0 2 1
0.496291 : 0 2 -1
Further details of the SOC calculation such as the procedure of SOC integral calculation can be controlled
via the %rel block (section 9.14.
9.19.2 Transition Metal L-Edges with ROCIS or DFT/ROCIS
The orca rocis program was designed to calculate transition metal L-edge spectra of large molecules as they
are observed in X-ray absorption spectroscopy (XAS). An L-edge results when an electron is promoted from
the 2p shell of a transition metal ion into the valence d shell by an X-ray photon. Strong spin-orbit coupling
in the 2p shell and p-d coupling phenomena complicate the interpretation and even more so the prediction of
these spectra. It has to be kept in mind that the present program applies a variety of approximations which
might lead to observable deviations from experimentally determined spectra. However, we believe that the
results obtained from the program are in general qualitatively correct and in most cases accurate close to the
experimental uncertainty. In cases where quantitative accuracy is not met, the provided results might still
give some insight into the mechanisms of intensity distribution in the spectra.
The special input structure for orbital windows described in 9.19.1 allows the user to restrict the donor
orbital space to the transition metal 2p shell. The acceptor orbital space is the same as in regular UV/Vis
spectroscopy. It should include all singly occupied molecular orbitals and as many virtual orbitals as one can
afford in the calculation. The number of roots should be chosen large enough so that at least all 2p-3d single
excitations are calculated. In many cases even more roots are required since doubly excited or charge transfer
states may become important. Moreover the strong SOC apparent in the 2p shell of transition metal ions
necessitates the additional calculation of excited states with a total spin of S 0 = S + 1 and S 0 = S 1 where
S is the total spin of the electronic ground state. Accordingly four additional excitation classes introduce
excited configuration state functions with a lower and higher spin multiplicity. They feature the second
quantized spin raising and lowering operators Spq+
=a p , Spq
q a
q a
=a p .
SOMO
E q
(t) 2S 0 +1
P 1 1 t
i = 2S 0 +2 Sti |0i S E |0i
2S 0 +1 2S 0 +2 uu i

u6=t

SOMO
E q

(t) 2S 0 +1
P 1 1 t

i = S
2S 0 +2 ti |0i S
2S 0 +1 2S 0 +2 uu i
E |0i

E q
u6=t
SOMO q S0 = S 1
(a) 2S 0 +1
= 2S 0 +3 Sai |0i
P (S 0 +1)2 S 02
1 a
S E |0i

(9.254)
(S 0 +1)(2S 0 +3) 2(2S 0 +2) tt i
i
t

SOMO

P q 2
q
1 +

+ S S S |0i

(2S 0 +2)(2S 0 +3) (2S 0 +2)2(2S 0 +1) tt uu ai
t,u6=t +E o
+
|0i S 0 = S + 1
a
i = Sai
Inclusion of configuration state functions with higher or lower multiplicity is invoked with the keywords
DoLowerMult and DoHigherMult, respectively.
%rocis NRoots 20
SOC true
DoRI true
PrintLevel 3
DoLowerMult true #Invokes a CI calculation #with S=S-1
DoHigherMult true #Invokes a CI calculation #with S=S+1
OrbWin = 6,8,0,2000
end
The program will conduct a separate Davidson procedure for each multiplicity. Subsequently it gives the
excitation energies and compositions of the calculated excited states for all included multiplicities. After all
CI calculations are finished, the program gives a list of all calculated roots with their excitation energies and
their multiplicities. It is this number that will be referred to as label I in the decomposition of spin-orbit

coupled states in the basis SMI . It is very important to note, that when states with different multiplicities
are calculated this number might deviate from the number that appears in the respective CI part of the
output. If one gets confused about the numbering of the states, the state energies might act as a guideline
through the output of the program.
Without SOC the spin exclusion rule applies which means that only excited states with a total spin equal to
the ground state spin (S 0 = S) give rise to non-vanishing intensities. Hence, only these transitions are listed
in the spectra before SOC.
--------------------------------------------------------------------------------
ROOT Mult Excitation energy[Eh] [cm-1] [eV]
--------------------------------------------------------------------------------
0 5 0.00000000 0.00 0.000
1 5 26.24822856 5760820.28 714.251
2 5 26.24833619 5760843.90 714.254
3 5 26.27159871 5765949.43 714.887
4 5 26.27982129 5767754.08 715.110
5 5 26.30321870 5772889.22 715.747
6 5 26.30458669 5773189.46 715.784
7 5 26.33143414 5779081.79 716.515
8 5 26.33600432 5780084.83 716.639
9 5 26.33865219 5780665.97 716.711
10 5 26.34522494 5782108.52 716.890
11 5 26.34577552 5782229.36 716.905
12 5 26.35183534 5783559.34 717.070
13 3 26.42121780 5798787.03 718.958
14 3 26.42122881 5798789.45 718.958
...
42 7 27.22926558 5976133.02 740.946

43 7 27.23201078 5976735.52 741.021
44 7 27.23280499 5976909.83 741.042
45 7 27.23594814 5977599.67 741.128
46 7 27.23865050 5978192.77 741.201
47 7 27.26590445 5984174.32 741.943
48 7 27.26597947 5984190.78 741.945
49 7 27.26604364 5984204.87 741.947
50 3 27.29447169 5990444.10 742.720
51 3 27.30121861 5991924.88 742.904
52 3 27.30655497 5993096.08 743.049
53 3 27.30685328 5993161.55 743.057
54 3 27.31274496 5994454.62 743.218
55 7 27.52164817 6040303.58 748.902
56 7 27.52433114 6040892.42 748.975
57 7 27.52448641 6040926.50 748.979
58 7 27.53903479 6044119.50 749.375
59 7 27.53935644 6044190.10 749.384
------------------------
ROCIS-EXCITATION SPECTRA
------------------------
NOTE: At this point no SOC is included!!!

Hence only transitions to states with the same spin multiplicity
as the ground state are observed!!!
Center of mass = ( -0.0011, -0.0021, 0.0000)

Calculating the Dipole integrals ... done
Calculating the Linear Momentum integrals ... done
-----------------------------------------------------------------------------
-----------------------------------------------------------------------------
(cm-1) (nm) (au**2) (au) (au) (au)
-----------------------------------------------------------------------------
1 5760820.3 1.7 0.000985130 0.00006 0.00612 -0.00434 0.00011
2 5760843.9 1.7 0.000777158 0.00004 -0.00008 0.00006 0.00666
3 5765949.4 1.7 0.000000036 0.00000 0.00000 0.00001 -0.00004
4 5767754.1 1.7 0.000007564 0.00000 0.00033 0.00057 -0.00000
5 5772889.2 1.7 0.025379335 0.00145 -0.00031 0.00021 -0.03804
6 5773189.5 1.7 0.026898175 0.00153 0.03203 -0.02254 -0.00039
7 5779081.8 1.7 0.000000323 0.00000 -0.00006 -0.00009 -0.00008
8 5780084.8 1.7 0.001711738 0.00010 -0.00572 -0.00805 0.00001
9 5780666.0 1.7 0.113054940 0.00644 -0.04616 -0.06564 -0.00001
10 5782108.5 1.7 0.151287595 0.00861 0.00073 -0.00052 0.09281
11 5782229.4 1.7 0.147199895 0.00838 0.07488 -0.05266 -0.00088
12 5783559.3 1.7 0.000000026 0.00000 0.00001 -0.00001 0.00004
28 5960986.7 1.7 0.004292708 0.00024 -0.00881 -0.01263 -0.00000
29 5963084.1 1.7 0.001638281 0.00009 -0.00774 0.00553 0.00006
30 5963136.7 1.7 0.001369356 0.00008 -0.00005 0.00003 -0.00869
31 5963484.9 1.7 0.000935993 0.00005 0.00415 0.00587 -0.00000
32 5968477.0 1.7 0.000661255 0.00004 0.00493 -0.00349 -0.00007
33 5968705.6 1.7 0.000607238 0.00003 0.00006 -0.00004 0.00579
35 5970943.7 1.7 0.000000001 0.00000 0.00000 0.00000 -0.00001
After calculation of SOC in the basis of all calculated ROCIS roots, the program prints out the composition
of the spin-orbit coupled states (if PrintLevel >2) and the corresponding absorption spectrum.
Eigenvectors of SOC calculation:

the threshold for printing is: 0.010000
weight : Root Spin Ms
State 0: 0.00 cm**-1 0.00000 eV
0.129027 : 0 2 2
0.741116 : 0 2 0
0.129027 : 0 2 -2
-------------------------------------------------------------------------------
SPIN ORBIT CORRECTED ABSORPTION SPECTRUM VIA TRANSITION ELECTRIC DIPOLE MOMENTS
-------------------------------------------------------------------------------
(cm-1) (nm) (au**2) (au) (au) (au)
-------------------------------------------------------------------------------
0 1 0.0 0.0 0.000000000 0.00000 0.00000 0.00000 0.00000
0 2 0.8 0.0 0.000000000 0.00000 0.00000 0.00000 0.00000
0 3 0.8 0.0 0.000000000 0.00000 0.00000 0.00000 0.00000
0 4 1.0 0.0 0.000000000 0.00000 0.00000 0.00000 0.00000
0 5 5729330.4 1.7 0.000080556 0.00002 0.00013 0.00009 0.00464
0 6 5729330.4 1.7 0.000096984 0.00003 0.00415 0.00295 0.00013
0 7 5731365.3 1.7 0.000000001 0.00000 0.00001 0.00000 0.00000
0 8 5731365.4 1.7 0.000000000 0.00000 0.00000 0.00000 0.00001
0 9 5733452.5 1.7 0.000058329 0.00002 0.00323 0.00227 0.00004
0 10 5733477.2 1.7 0.000066389 0.00002 0.00003 0.00002 0.00421

0 11 5734964.4 1.7 0.000000034 0.00000 0.00005 0.00007 0.00004
0 12 5737151.2 1.7 0.000047769 0.00001 0.00208 0.00291 0.00000
With the aid of the orca mapspc program it is possible to extract a .plt file from the printed spectra, which
then can be used to generate a plot of the intensity vs the excitation energy. The orca mapspc program
applies Gaussian type lineshape functions to the calculated transitions with a user-defined FWHM. One has
to provide some information for the program such as the name of the output file, the type of spectrum you
wish to plot, the energy range and the like. It is invoked in the command line and the parameters are given
as arguments:
orca_mapspc FeIICl4.out socabs -eV -w1 -n3000 -x0710 -x1740
The first argument has to be the output file of your calculation followed by the type of spectrum that should
be plotted. In the case of transition metal L-edges it is an absorption spectrum after the SOC correction. The
arguments -eV (use electron Volt as energy unit), -w1 (FWHM = 1eV), -n3000 (use 3000 grid points),
-x0710 and -x1740 (energy range: 710 to 740 eV) have to be adapted to the specific calculation. As a
result, one obtains a .plt and a .stk file. The .plt file contains five columns. In the first column one finds
the energy and in the second the total intensity. Columns three to five contain the x-,y- and z-components of
the transition moment. Note, that the distribution of the transition moment among its spatial components
depends on the orientation of your molecular axis system. The .stk file contains a list of all transitions with
their respective transition energy and intensity. A more detailed description of the orca mapspc program
and its usage can be found in chapter 9.32.1.
Figure 9.24: Comparison of the experimentally observed (black) and calculated ROCIS (red) Fe
L-edge of [FeCl4 ]2 . The red bars highlight the contribution of individual states to
the total spectrum. The calculation was performed using the TZVP basis set.
For many transition metal compounds the description of the electronic ground and excited states by Hartree-
Fock theory and CIS is of rather poor quality. Especially covalency and relative spin state energetics are not
reproduced correctly. This in turn might lead to wrong intensity distributions in the calculated L-edge spectra.
In the majority of these cases the quality of the description and hence the predicted L-edge spectra can be
significantly improved with the DFT/ROCIS method [381]. It features the usage of a restricted open-shell
Kohn-Sham matrix as reference and also uses the DFT orbitals for setting up the excited configuration state
functions in the CI expansion. The two electron integrals that include the DFT orbitals are scaled according
to their nature and their position in the CI matrix by the parameters c1 , c2 and c3 . They all lie in the interval
[0;1]. Parameters c1 and c2 scale coulomb- and exchange- like terms in the diagonal part of the CI matrix,
whereas c3 reduces the size of all off-diagonal elements of the CI matrix. For example:
DFT/ROCIS C(KS) C(KS)

Hia,ia = Faan Fii c1 (ii|aa) + 2c2 (ia|ia)
DFT/ROCIS C(KS) C(KS)
o (9.255)
Hia,jb = c3 ij Fab ab Fji (ij|ab) + 2 (ia|jb)
The three default parameters c1 = 0.21, c2 = 0.49 and c3 = 0.29 have been optimized for a test set of
molecules and their excited states on a B3LYP/def2-TZVP(-f) level of theory but can be freely chosen [381].
It is most likely that for a different combination of test molecules, functional and basis set, a different set of
parameters gives better results. Since the parameters are chosen with regard of a good balance between
orbital energies, Coulomb and exchange integrals, a new set of parameters should at least crudely resemble
their relative proportions.
! B3LYP def2-TZVP(-f) def2-TZVP/C TightSCF Grid4 NoFinalGrid
%ROCIS NRoots 20
DoRI true
SOC true
DoHigherMult true
PrintLevel 3
OrbWin = 5,7,50,60
DoDFTCIS true #switches on the DFT/ROCIS method
DFTCIS_c = 0.21, 0.49, 0.29 #Array input of the three parameters
end
Figure 9.25: Comparison of the experimentally observed (black) and calculated (red) Ti L-edge of
[Cp2 TiCl2 ]. The red bars highlight the contribution of the individual states to the
total spectrum. The pure ROCIS method (left) predicts a wrong L3 -L2 intensity ratio
and strongly overestimates the splitting of the satellite features to the main bands.
Better results are obtained with the DFT/ROCIS method (right).
9.19.3 Natural Transition Orbitals/ Natural Difference Orbitals
Likewise to CIS and TD-DFT (section 9.18.7) The nature of the calculated excited states in ROCIS and
DFT/ROCIS can be analyzed by using the Natural Transition Orbitals (NTO) or Natural Difference Orbitals
(NDO) machineries. [382] Note that:
The NTO analysis is based on the transition density between ground and excited states. Hence is valid
for singly excited states and for states of the same multiplicity.
The NDO analysis on the otherhand is somewhat more flexible in this respect as it is based on the
difference density between ground and excited states.
Presently, only one analysis (NTO or NDO) can be performed at a time while when both flags are on
the NTO analysis switches off.
An example is given below for [FeCl4 ]2 :
!B3LYP def2-TZVP def2/J Conv TightSCF LargePrint PAL4
%ROCIS NRoots 40
PrintLevel 3
MaxCore 4000
MaxDim 360
SOC true
DoRI true
DoNTO true
DoNDO true
NDOThresh/NTOThresh 1e-4
NDOStates/NTOStates= 1,2,3,4,5,6,7,8,9,10,13,14,15
DoLowerMult true
DoHigherMult true
DoDFTCIS true
DFTCIS_c = 0.18, 0.20, 0.40
OrbWin = 6,8,0,2000
end
* xyz -2 5
Fe -17.84299991694815 -0.53096694321123 6.09104775508499
Cl -19.84288422845700 0.31089495619796 7.04101319789001
Cl -17.84298666758073 0.11868125024595 3.81067954087770
Cl -17.84301352218429 -2.87052442818457 6.45826391412877
Cl -15.84311566482982 0.31091516495189 7.04099559201853
*
Then the respective NTO and NDO analysis for state 15 is given below:
------------------------------------------
------------------------------------------
done
Solving eigenvalue problem for the Occupied space ... done
Solving eigenvalue problem for the Acceptor space ... done
Natural Transition Orbitals were saved in nto.14.nto
Threshold for printing occupation numbers 1.0e-04
E= 25.447756 au 692.469 eV 5585137.0 cm**-1

49[0] -> 46[1] : n= 0.39056909
48[0] -> 47[1] : n= 0.08619374
47[0] -> 48[1] : n= 0.00441125
-------------------------------------------------
NATURAL DIFFERENCE ORBITALS FOR STATE 14
-----------------------------------------------
done
Solving eigenvalue problem for the Occupied space ... done
Solving eigenvalue problem for the Acceptor space ... done
Natural Difference Orbitals were saved in ndo.14.ndo
Threshold for printing occupation numbers 1.0e-04
E= 25.447756 au 692.469 eV 5585137.0 cm**-1

49[0] -> 46[1] : n= 0.81173217
48[0] -> 47[1] : n= 0.17903699
47[0] -> 48[1] : n= 0.01165859
46[0] -> 49[1] : n= 0.00922738
45[0] -> 50[1] : n= 0.00112567
For closed shell cases the orbitals are save in similar way to TDDFT and CIS (section 9.18.7). In the case of
open shell cases for convenience donor orbitals are saved with orbital operator 0 while acceptor orbitals with
orbital operator 1. This needs to be specified in the orca plot program and should not be confused with the
spin-up and spin-down orbitals in the UHF and UKS cases.
In practice one can use this machinery to analyze for example the relativistically corrected states located at
705.5 eV (when shifted with respect to experiment). It can be seen that these states contain for example
significant contributions from state 14. NTO or NDO analysis then shows that this state is dominated by the
spin conserving DOMO-SOMO 2pz 3dyz single electron excitation.
9.19.4 Resonant Inelastic Scattering Spectroscopy
9.19.4.1 General
Starting from ORCA version 4.0 ROCIS module can be used to calculate RIXS spectra
State
NDO49[0] NDO46[1]
Figure 9.26: DFT/ROCIS calculated L3 XAS spectrum of [Fe(Cl)4 ]2 together with NDO analysis
for state 14. Constant broadening 0.5 eV and isovalue for the orbital plots 0.03 a.u. is
used throughout
The present implementation is directly based on the Kramers Heisenerg Dirac (KDH) expression formula for
near resonant and resonant conditions
2 ( )
2
X X hF | m |V i hV | m |Ii F
| (Eex , Esc )|T otal = (9.256)
EV I Eex i 12 V 2
(EF V Eex + Esc ) + 41 F 2

F V
2
X 2 2
| (Eex , Esc , V )|resonant = |hF | m |V i| |hV | m |Ii| f (EV I , EF V , Eex , Esc , V , F ) (9.257)
F
2
X 2
| (EV I , Esc )|Direct = | (EV I , Esc , V )|resonant (9.258)
V
The resonance scattering cross section for total and direct cases, averaged over all orientations of the molecule
and integrated over all directions and polarizations of scattered radiation is given in equations:
3
T otal 8Esc Eex X 2
RXES (Eex , Esc ) = | (Eex , Esc )|T otal (9.259)
9c4
,=x,y,z
3
Direct 8Esc Eex X 2
RXES (Eex , Esc ) = | (Eex , Esc )|Direct (9.260)
9c4
,=x,y,z
Interference effects can be then derived in a straightforward way from equation:
interf erence T otal Direct

RXES (Eex , Esc ) = RXES (Eex , Esc ) RXES (Eex , Esc ) (9.261)
In order to access RIXS spectroscopy in the ROCIS module one needs in addition to specify a 2nd donor
space. This is specified by defining an OrbWin array with 6 elements: The first four elements define the
ranges of the two donor spaces while the last two elements the respective acceptor space range.
OrbWin 0,0,2,4,45,60
An important difference with respect to the conventional ROCIS or DFT/ROCIS calculations is the fact that
two donor spaces of very different energy ranges are involved (e.g. K-edge, L-edge) which requires to restrict
somewhat the acceptor space and saturate it with as many states as possible.
The main calling commands in order to perform a RIXS calculation within both ROCIS and CASSCF blocks
are the following:
RIXS true. Similar to absorption spectroscopy, this requests the RIXS calculation to be performed
based on the calculated non-relativistic ground state multiplicity States
RIXSSOC true. By turning-on this flag the RIXS is calculated by taking in account the relativistically
corrected Ms States.
Elastic true. This flag indicates whether the resonant condition in which the initial and Final states
coincide should be taken into account. Note that the intensity of this spectral feature might be
overestimated as presently the non resonant terms are not treated
The respective ROCIS input reads then as follows:
!B3LYP def2-TZVP def2/J SlowConv Grid5 NoFinalGrid
%ROCIS
NRoots 200
PrintLevel 3
MaxCore 4000
DoRI true
DoHigherMult true
SOC true
RIXS true #Request RIXS calculation (NoSOC)
RIXSSOC true #Request RIXS calculation (with SOC)
Elastic true #Request RIXS calculation (Elastic)
DoDFTCIS true
DFTCIS_c =0.18,0.20,0.40
OrbWin = 2,4,25,33,0,100
end
* xyzfile 2 2 test.xyz
When running the calculation one can monitor if the requested NRoots were sufficient enough to select the
states dominated by both the donor orbital spaces
--------------------------------------------------------------------------------
ROOT Mult Excitation energy[Eh] [cm-1] [eV]
--------------------------------------------------------------------------------
0 2 0.00000000 0.00 0.000
1 2 0.06611737 14511.08 1.799
2 2 0.07728471 16962.03 2.103
3 2 0.07732428 16970.72 2.104
...
84 2 33.75471831 7408304.35 918.513
85 2 33.77073325 7411819.22 918.948
86 2 33.77076955 7411827.19 918.949
87 4 34.06882971 7477243.83 927.060
88 2 34.07021441 7477547.74 927.098
...
If that is not the case the respective RIXS calculations will not be performed and a Warning Message will be
generated:
Making the RIXS files ...

WARNING!: Flag for RIXS property calculation was identified but
there is zero number of Intermediate and/or Final states:
No Cross-Section properties will be evaluated ...Skipping this part
TIP: Increase the number of NRoots and/or decrease or increase
the acceptor orbital space
...Done
A successful run on the other hand will generate the following messages for RIXS and RIXSSOC calculations.
----------------------------------------------------------------------------------
ROCIS RIXS SPECTRUM
----------------------------------------------------------------------------------
Making the RIXS data files for Inelastic and Elastic Scattering
Ground State: 1
Intermediate States: 21
Final States: 59
The RIXS cross section will be generated from:
60 Ground-Final State Pairs and 21 Intermediate States/Pair
Calculating Intensities...
10% done
20% done
30% done
40% done
50% done
60% done
70% done
80% done
90% done
100% done
Storing the files...All Done
----------------------------------------------------------------------------------
----------------------------------------------------------------------------------
ROCIS RIXSSOC SPECTRUM
----------------------------------------------------------------------------------
Making the RIXS-SOC data files for Inelastic and Elastic Scattering
Ms States: 2
Intermediate States: 78
Final States: 214
The RIXS cross section will be generated from:
432 Ground-Final State Pairs and 78 Intermediate States/Pair
Calculating Intensities...
10% done
20% done
30% done
40% done
50% done
60% done
70% done
80% done
90% done
100% done
Storing the files...All Done

----------------------------------------------------------------------------------
In both cases the number of involved Initial, Final and Intermediate states is specified explicitly.
For example in the case of RIXSSOC 2 Ms Ground states, 78 Intermediate states and 214 Final states are
involved. Then the RIXS cross section for elastic and inelastic scattering will be generated by 432 (2*(2+214))
Ground-Final State-Pairs and 78 Intermediate States per Ground-Final state pair.
9.19.4.2 Processing the spectra with orca mapspc
By calling orca mapspc with the following keywords:
orca_mapspc test.el_inel.rocis.rixssoc RIXS -x0871 -x1876 -x2-1 -x34 -w0.4 -g0.4

-l -n125 -m125 -dx20 -eaxis1
The program will process the test.el inel.rocis.rixssoc file with the following parameters:
Energy range along x : 871-876 eV
Energy range along y: -1-34 eV

-l indicates Lorentzian broadening
Width along x (gamma): 0.4 eV
Width along y (gamma): 0.4 eV
Points along x: 125
Points along y:125
Shift to be applied along Incident energy/Emission axis: 20 eV
The y axis will be Energy Transfer axis. If -eaxis2 is the y axis will be then Emission Energy axis
All this information is printed during the data processing:
Mode is RIXS
Using Lorentzian shape
Cannot read the paras.inp file ...
taking the line width parameter from the command line
Cannot read the udex.inp file ...
taking the excitation energy ranges from the command line
Cannot read the udem.inp file ...
taking the emission energy ranges from the command line
Cannot read the gfsp.inp file ...
No Ground-Final State Pairs will be evaluated
---------------------------------------------------------------------------------
PLOTTING RIXS SPECTRA
---------------------------------------------------------------------------------
Input File : test.el_inel.rocis.rixssoc
Incident Energy Excitation axis : 871.000 ... 876.000 eV 125 points
Energy transfer axis : -1.000 ... 4.000 eV 125 points
Incident Energy Shift : 20.000 eV
Lorenzian Linewidth along Incident Axis : 0.400 eV
Lorenzian Linewidth along Energy Transfer/Emission Axis : 0.400 eV
y axis : 1 -> Energy transfer
Number of user defined cuts at constant Excitation Energy axis: 0
Number of user defined cuts at constant Emission/Energy Transfer Energy axis : 0
Making checks...Done
Proccessing data...
10% done
20% done
...
100% done
RIXS-plotting done
Incident Energy range: 845.800 ... 869.249
Emission/Energy-transfer range: 0.000 ... 4.853
Now storing the 2D file...

Done
Making the Integrated spectra along Energy Transfer/Emission axis... Done

Making the Integrated spectra along Incident axis... Done
All Done
---------------------------------------------------------------------------------
Successful run will generate the following files: The RIXS planes of the Total, Direct and Interference RIXS
intensity as indicated in the above equations:
test.el_inel.rocis.rixssoc.total_rixs.dat
test.el_inel.rocis.rixssoc.direct_rixs.dat
test.el_inel.rocis.rixssoc.interference_rixs.dat
In addition one obtains the integrated spectra at constant Incident energies (CIE):
test.el_inel.rocis.rixssoc.dw.dat
as well as at constant Emission/Energy Transfer energies (CEE/CET):
test.el_inel.rocis.rixssoc.wex.dat
Figure 9.27: DFT/ROCIS calculated RIXS planes for [Cu(N H3 )4 ]2 . Left: Total RIXS Intensity,
Middle: Direct RIXS intensity and Right: Interference RIXS intensity. Lorentzian
lineshape broadening with constant widths along Incident and Energy Transfer axis
(0.5 and 0.2 eV respectively) were used throughout.
9.19.4.3 Generating Cuts
Cuts along x and y axis can be generated with two ways:
1) At first, this action can be performed by adding the following keywords: uex and udw accounting for
generating cuts at constant Incident Energies (CIE) and at constant Emission (CEE)/or at constant Energy
Transfer (CET) respectively, together with the desired number of cuts.
2) Alternatively, the energies of the desired cuts can be specified as lists in the files udex.inp (user defined
excitations) udem.inp (user defined emissions)
For example if in udex.inp one specifies:
872.5
874.2
and for the cuts along Energy Transfer axis one just specify -udw3
orca_mapspc test.el_inel.rocis.rixssoc RIXS -x0871 -x1876 -x2-1 -x34 -w0.4 -g0.4

-l -n125 -m125 -dx20 -eaxis1 -udw3
Then at the end one gets:
Making the specified cuts (2) at constant Excitation Energy axis...

Writing file: test.el_inel.rocis.rixssoc_872.50.rxes_vs.dat ...Done
Writing file: test.el_inel.rocis.rixssoc_872.50.rxes_fs.dat ...Done
Writing file: test.el_inel.rocis.rixssoc_874.20.rxes_vs.dat ...Done
Writing file: test.el_inel.rocis.rixssoc_874.20.rxes_fs.dat ...Done
Done
Making the specified cuts (3) at constant Emission/Energy Transfer axis...

Writing file: test.el_inel.rocis.rixssoc_-1.00.xas_vs.dat ...Done
Writing file: test.el_inel.rocis.rixssoc_-1.00.xas_fs.dat ...Done
Writing file: test.el_inel.rocis.rixssoc_1.50.xas_vs.dat ...Done
Writing file: test.el_inel.rocis.rixssoc_1.50.xas_fs.dat ...Done
Writing file: test.el_inel.rocis.rixssoc_4.00.xas_vs.dat ...Done
Writing file: test.el_inel.rocis.rixssoc_4.00.xas_fs.dat ...Done
Done
All Done
---------------------------------------------------------------------------------
The files * rxes fs.dat are RXES spectra containing all individual contributions from all Final states together
with the Direct, the Total and the Interference contributions at the given constant Incident Energy.
Similarly, the * rxes vs.dat are RXES spectra containing individual contributions of the Intermediate states,
together with the Direct the Total and the Interference contributions at the given constant Incident Energy
Likewise, the respective * xas fs.dat and * xas vs.dat are XAS type spectra with individual contributions
at a given constant Emission or Energy transfer Energy
These files are Energy vs Intensity files and read like:
1) for *fs.dat
X S- 1( 0- 0) S- 2( 0- 1) DIRECT TOT INTERFERENCE
2) for *vs.dat
X S- 1( 45) S- 2( 47) DIRECT TOT INTERFERENCE
In the first case S -1(0-0) represents the individual contribution of a given Ground-Final state pair. The
numbering follows the numbering of the output file e.g.:
Eigenvalues: cm-1 eV Boltzmann populations at T = 300.000 K

0: 0.0000 0.0000 3.44e-01
1: 8.3818 0.0010 3.31e-01
Hence, in this case S -1 represents the elastic scattering intensity.
In the second case S -1(45) represents the individual contribution of a given Intermediate state.
44: 66918.6071 8.2968 1.43e-140

45: 6996678.8061 867.4775 0.00e+00
46: 6996693.0276 867.4793 0.00e+00
In this case S -1 represents the intensity contribution of the first Intermediate state.
9.19.5 Keyword List
%rocis
#-----------------------------------------------------------
# GENERAL KEYWORDS
#-----------------------------------------------------------
NRoots 3 #The number of desired roots
MaxDim 5 #Davidson expansion space = MaxDim * NRoots
MaxIter 35 #Maximum CI Iterations
NGuessMat 512 #The dimension of the guess matrix
ETol 1e-6 #Energy convergence tolerance
RTol 1e-6 #Residual Convergence tolerance
MaxCore 2000 #Maximum memory used during the calculation in MB
EWin= -5,5,-5,5 #Energy Window that defines orbital excitation space
OrbWin=6,8,0,2000 #Orbital Window that defines orbital excitation space
#(overrides EWin)
DoRI false #Switch for the RI approximation
DoLoc false #Switch for localization of Donor orbital space
LocMet PipekMezey #chooses the localization method:
#PipekMezey or FosterBoys.
#Abbreviations "PM" and "FB"

#are equivalent to full names.
SOC false #Switch for inclusion of SOC
SOCTemp 10 #The fictive temperature for the
#SOC corrected spectra
DoDFTCIS false #Switch for the DFT/ROCIS method
DFTCIS_C = 0.18, 0.20, 0.40 #Array Input of the
#three DFT/ROCIS parameters
#-----------------------------------------------------------
# FLAGS FOR EXCITATION SPACES
#-----------------------------------------------------------
Do_is true #Include DOMO->SOMO excitations
Do_sa true #Include SOMO->Virtual excitation
Do_ia true #Include DOMO->Virtual excitations
Do_ista true #Include DOMO->SOMO excitations
#coupled to SOMO->Virtual
#excitations with s not equal t
Do_isa true #Include DOMO->SOMO excitations
#coupled to SOMO->Virtual
#excitations with s = t
DoLowerMult false #Switch for excitation with S=S-1

Do_LM_is true #Include DOMO->SOMO excitations
#with S=S-1
Do_LM_sa true #Include SOMO->Virtual excitations
#with S=S-1
Do_LM_ia true #Include DOMO->Virtual excitations
#with S=S-1
DoHigherMult false #Switch for DOMO->Virtual

#excitations with S=S+1
#-----------------------------------------------------------
OUTPUT KEYWORDS
#-----------------------------------------------------------
PrintLevel 3 #Controls the amount of output
#produced during the calculation
DoCD false #Invokes the calculation of CD spectra
DoQuad false #Invokes the calculation of
#electric quadrupole, magnetic dipole
#magnetic quadrupole and electric octupole
#contributions to the calculated spectra
RIXS false #Perform a RIXS calculation
RIXSSOC false #Perform a RIXS calculation on the basis
#of relativistically corrected states
Elastic false #Include the elastic line in the generation
#of the RIXS or RIXSSOC spectra
PlotDiffDens = 1,2 #Array input for plotting
#difference densities of CI roots
#1 and 2 to the ground state.
PlotSOCDiffDens=1,2 #Array input for plotting
#difference densities of SOC
#states 1 and 2 to the ground state
DoNTO false #Request Natural Transition Orbital Analysis
DoNDO false #Request Natural Difference Orbital Analysis
9.20 Excited States via EOM-CCSD 545
#(if true it switches off the NTO analysis)

NDOThresh 1e-4 #Threshold for printing occupation numbers
NTOThresh 1e-4 #Threshold for printing occupation numbers
NDOStates = 1,2 #Array input for states to be taken into account
NTOStates = 1,2 #Array input for states to be taken into account
Weight_Thresh 0.01 #Threshold for contributions to CI
#and SOC states to be printed
9.20 Excited States via EOM-CCSD
The EOM-CCSD routine is part of the orca mdci module of the ORCA program package. It is called after a
successful coupled-cluster calculation, if the appropriate flags and the number of roots have been set. In the
following chapter the general program flow and all input parameters of the EOM routine will be described in
detail (for typical use, see 8.7). For an RHF reference, the EE-, IP- and EA-EOM-CCSD approaches are
available for the computation of excitation energies, ionization potentials and electron affinities of closed-shell
molecules, respectively. For a UHF reference, only the UHF EE-EOM-CCSD approach is currently available
for the calculation of excitation energies in open-shell molecules.
The EOM wave function is parametrized in the following manner
R|CC i, (9.262)
i.e. via the action of a linear excitation operator R on the coupled-cluster ground state wave function CC .
Here, R is a particle conserving operator, in the case of the excitation energy problem. However, this is
not true for the ionization or electron attachment problems, where R is a net annihilation or net creation
operator, respectively. The ground state coupled-cluster T-amplitudes are obtained from a CCSD calculation,
and our task is to obtain R. Note that since the CC Hamiltonian is nonsymmetric, a left hand solution
(L) would also be needed to evaluate properties. For the calculation of excitation, ionization or electron
attachment energies, however, it is enough to obtain the right hand solutions (R). In principle, this is done
by building the Hamiltonian and diagonalizing it in order to obtain energy expectation values.
In practice, the size of the CCSD Hamiltonian matrix is prohibitively large and thus, various methods have
been devised to obtain its lowest few eigenvalues and eigenstates. One of the most popular of these approaches
is the Davidson method, which can be summarized as follows:
Construct an initial guess of orthogonal trial vectors, C.
Evaluate the sigma vectors = HC.
Build model Hamiltonian H = C T .
Diagonalize H: E = U T HU.
Compute Ritz vectors X = CU.

Compute residuals R = XE U, check convergence: if yes, pass X, E as solutions.
Preconditioning: T = M R (many possible choices for the preconditioner M ).
Check if adding new trial vectors would exceed the maximum number of trial vectors:
if no, add T to C, and orthonormalize the united set
if yes, then set X as C (orthonormalize if H is nonsymmetric); then add T and orthonormalize
The advantage of the above method is that, instead of the full Hamiltonian, only the sigma vectors have to
be explicitly evaluated and stored.
It is also possible to use a lower scaling version of the EOM-CCSD methods, which relies on the perturbative
truncation of the coupled-cluster similarity transformed Hamiltonian. Presently, only the second order
truncated version (CCSD(2) approximation) is available. However, it is better to use the PNO based
implementation, as it has the cost of EOM-CCSD(2), but its accuracy is comparable to canonical EOM-CCSD.
Note that EOM-CCSD(2) and the PNO based implementation are currently only available for closed-shell
molecules within the RHF based implementation.
Below are all the parameters that influence the RHF EOM routine. In the following sections, these parameters
will be explained following the solver algorithm described above.
%mdci
#EOM parameters - defaults displayed
DoEOM false # whether to perform EOM
UseEOMOptS true # use optimized sigma routines for singles
UseEOMOptD true # use optimized sigma routines for doubles
NDav 20 # maximum size of reduced space (i.e. 20*NRoots)
CCSD2 false # Use the lower scaling CCSD(2) approximation
CheckEachRoot true # check convergence for each root separately
RootHoming true # apply root homing
DoLanczos false # use the Lanczos procedure rather than Davidson
UseCISUpdate true # use diagonal CIS for updating
NInitS 0 # number of roots in the initial guess, if 0, use preset value
DRESS3ES true # construct the external dressing to singles
# or calculate on the fly
DRESS3ED false # construct the external dressing to doubles
#or calculate on the fly
DOCOSXEOM false # use COSX approximation for external exchange term in EOM
DOAOX3E false # use COSX approximation for 4 external terms contribution
# to 3 external intermediate
DoRootwise false # solves for each root separately,
# more stable for large number of roots
DoTDM false # option for calculation of default transition moment
Doleft false # calculation of exact left vector
NRootsPerBatch 1 # no of roots calculated together
FOLLOWCIS false # follows the initial singles guess
DoCore true # initiates ionization or excitation from core orbital

CoreHole 0 # core orbital from which ionization or excitation is needed
CVSEP false # separates core orbital from valence,
DTol 1e-5 # default for EOM residual threshold
#keywords which affect EOM parameters, but do not belong to the routine itself
NRoots 9 # number of roots
OTol 1e-14 # orthogonalization threshold
KCOpt KC_MO # method for external exchange formation
KC_AOX # when asked for exact TDM calculation
KC_AOBLAS # most efficient
PrintLevel 3 # the amount of information to be printed
MaxCore 500 # total amount of memory
end
In the case of the UHF EE-EOM-CCSD implementation, the parameters that influence a given calculation
are provided below.
%mdci
#UHF EOM parameters - defaults displayed
DoEOM false # whether to perform EOM
NDav 20 # maximum size of reduced space (i.e. 20*NRoots)
CheckEachRoot true # check convergence for each root separately
RootHoming true # apply root homing
NInitS 0 # number of roots in the initial guess, if 0, use preset value
DoRootwise true # solves for each root separately (default in UHF case)
FOLLOWCIS true # follows the initial singles guess
DTol 1e-5 # default for EOM residual threshold
#keywords which affect EOM parameters, but do not belong to the routine itself
NRoots 9 # number of roots
OTol 1e-14 # orthogonalization threshold
KCOpt KC_AOX # AO exchange for the four external contributions
# (the only option available at present)
PrintLevel 3 # the amount of information to be printed
MaxCore 500 # total amount of memory
end
9.20.2 Memory Management
The most important data coming from the coupled-cluster routine are the ground state energy and wave
function, and the molecular integrals. The integrals are then used to create dressed integral containers,
which allows for an efficient factorization of the EOM equations, since these dressed quantities do not change
during the calculation. Most of these are written on disk, with the possible exception of the integral container
which has three external labels. This, and the solver files may remain in core if enough memory is available.
The program sequentially tries to allocate memory for the files in the order of their importance, and what
cannot be kept in core, goes on disk. The order of allocation is as follows: 1. residual vectors, 2. Ritz vectors,
3. three external integrals, 4. sigma vectors and 5. state (trial) vectors, as seen in the example below:
--------------------------------
AUTOMATIC CHOICE OF INCORE LEVEL
--------------------------------
Memory available ... 6512.00 MB

Memory needed for Residual-vectors ... 71.27 MB
Memory needed for Ritz-vectors ... 71.27 MB
Memory needed for 3-ext integrals ... 92.05 MB
Memory needed for Sigma-vectors ... 1425.31 MB
Memory needed for State-vectors ... 1425.31 MB
-> Final InCoreLevel ... 5
Half of the memory specified with the keyword MaxCore is distributed among the five candidates. In the above
case, everything fits in memory. Note that these are only the largest contributors to memory consumption,
and there should ideally be a safety margin when allocating memory.
In order to estimate the amount of necessary memory, it should be kept in mind that, in the closed shell case,
the memory requirements of the residual and Ritz vectors are proportional to NR NP NV2 , the three external
integrals to NR NO NV3 and the sigma and trial vectors to ND NR NP NV2 , where NO and NV are the number
of occupied and virtual orbitals, NP = NO (N2O +1) is the number of occupied pairs, NR is the number of roots,
and ND is the maximum size of the reduced space. The keyword NRoots sets NR , while NDav determines
ND . Luckily, the contributions that, in our experience, are the most important to keep in memory, are also
the ones that require the smallest amount of it. It is advisable to use KCOpt AOBLAS, as it has the lowest
memory requirements.
Note that in the UHF EE-EOM-CCSD implementation, the memory requirements of the residual and Ritz
vectors are proportional to NR (NP NV2 + NP NV2 + NO NO NV NV ), the three external integrals to
NR (NO NV2 + NO NV2 + NO NV NV2 + NO NV NV2 ) and the sigma and trial vectors memory requirements
are proportional to ND NR (NP NV2 + NP NV2 + NO NO NV NV ), where NO , NO , NV and NV are
respectively, the number of occupied alpha, occupied beta, virtual alpha and virtual beta orbitals and
N (N 1) NO (NO 1)
NP = O 2O and NP = 2 are the number of alpha and beta occupied pairs, respectively.
9.20.3 Initial Guess
The present initial guess in the RHF EOM implementation consists of constructing a CIS Hamiltonian of a
certain dimension, and diagonalizing it. The roots are preselected based on the energetic ordering of the
diagonal elements of the Hamiltonian. In the UHF case, the guess is constructed from the solutions of a UHF
CIS calculation. The number of roots in the initial guess is determined as 20 times the number of roots desired
in EOM (NRoots) if NDav is 20 or smaller, otherwise it is set to NDav times the number of EOM roots. If the
parameter NInitS is larger than zero, then the number of initial guess roots will be set to this parameter
times NRoots. The maximum possible number of roots is the full CIS dimension, NO NV . One should keep in
mind, while increasing the number of initial guess vectors, that this corresponds to diagonalizing a matrix of
increasing dimension. If, for example NRoots is 10, then by default 200 roots are considered in the initial
guess (unless it exceeds the size of the CIS space), or if NInits is set to 100, then there will be 1000 roots in
the guess. In some cases, the roots calculated using EOM may not be the lowest ones, but a few of these may
be replaced by some higher roots which are easier to find. In such cases, it may help to increase NRoots or
NInitS to converge to the proper roots. The program can be made to follow the initial CIS guess by setting
FOLLOWCIS to true and is necessary if we wish to ionize or excite from inner-valence or core orbitals. In the
RHF implementation, the core orbital, from which the ionization or excitation originates, can be specified
using the keyword CoreHole, in addition to setting DoCore and FollowCIS to true. The CoreHole keyword is
quite general and in principle, ionization or excitation processes from any occupied orbital can be specified
using this keyword.
9.20.4 Hamiltonian Construction
The Hamiltonian construction begins by calling the sigma routines. The logical variables UseEOMOptS and
UseEOMOptD choose the routines to be used in the evaluation of the singles and doubles sigma vectors,
respectively. If true, the optimized sigma routine, using dressed integrals, will be used. This should not be
changed, the option is there mainly for debugging purposes. If set to false, an automatically generated, and
much slower serial code will be used instead. In each early iteration, NR sigma vectors will be determined,
except in thecase of a restart, where the number of sigma vectors is 2NR . For further details on convergence,
see 9.20.6 below.
The most time consuming part of the sigma vector construction is the formation of the external exchange
contribution, which can be influenced via the CC keyword KCOpt. Currently, there are three options that are
compatible with EOM: KC MO,KC AOX and KC AOBLAS (see the MDCI documentation). The external exchange
term can be treated most efficiently using COSX, which leads to average speed ups of 10x for the external
exchange term and an overall speedup of 3x for the EOM calculation. This is accompanied by a drastic
reduction of the storage cost [383]. The error introduced is below 1 meV, which is 200-fold less than the error
bar of the method [383] itself. It is the default for KCOpt KC AOX and KC AOBLAS and can be controlled by the
keyword DOCOSXEOM. The default grid settings for EOM are GridX 1 and IntAccX 2.68.
Once the sigma vectors are available, they are multiplied with the trial vectors to yield the reduced space
Hamiltonian. The Hamiltonian is built in a way that, in each iteration, only the new vector products are
added to the edge of the old Hamiltonian, so that a full build is avoided. It should be clear that the
parameter NDav plays an important role here, since it determines the maximum size of the Hamiltonian
(ND NR ), and also controls how much memory is needed for the trial and sigma vectors, as seen above. Since
the choice of this parameter influences convergence properties, it will be discussed further in 9.20.6.
9.20.5 Solution of the (Nonsymmetric) Eigenproblem
Following the construction of the Hamiltonian, a nonsymmetric eigensolver is called. In this case, it is possible
to have complex eigenvalues. In practice, this is rarely the case, and indicates a problem of some kind. A
warning will be given if this happens, however, one may get away with this if it only happens in an isolated
iteration step.
Once the eigenvectors are available, they are compared with those of the previous iteration, if root homing is
turned on, i.e. if the RootHoming keyword is set to true. This means evaluating the overlap of the old and
new eigenvectors, in order to keep track of the possible movement of the eigenvectors if root flipping occurs.
If converged roots are removed from further iterations (see next section), it is important to keep track of
changes in ordering, especially if a converged and a non-converged root is swapped. After diagonalization,
the Ritz vectors and residuals can be evaluated.
9.20.6 Convergence, Restart, Preconditioning and Subspace Expansion
Convergence is signaled once a residual square norm based criteria is fulfilled. This criteria is determined
by the CheckEachRoot keyword. If it is true (default), the convergence of the residual square norm of each
root is checked separately. This is due to the fact that different roots converge at a different rate. Once a
root is converged, no new trial vectors will be generated, belonging to that vector. This means that the
EOM iterations will progressively become faster (until restart). Turning off the rootwise convergence check
is possible, but not recommended. In this case, the maximum of all residual square norms is checked for
convergence, and all iterations will take roughly the same amount of time since no vectors are removed in
any iteration. However, this procedure can be numerically unstable, since the residuals of some roots might
become very close to zero, and trying to generate new vectors, which are orthogonal to these, may lead to
numerical disaster. In short, the recommended default is having both CheckEachRoot and RootHoming set
to true. If CheckEachRoot is false, then RootHoming should also be set to false, as it may cause problems if
NDav is too small. The convergence threshold of the residual in Davidsons method can be largerthan that for
the ground state CC residual threshold in order to obtain converged results. Namely, a value of DTol of 1e-5
is almost always enough to get well converged energies.
At this point it is worth discussing the role of the keyword NDav. This keyword determines at what point
the Davidson algorithm should be restarted. If it is chosen too small, it may cause slow convergence. If this
value is too large, this may result in overwhelming demands on memory/disk space requirements. The default
value (20) is chosen with the hope that no, or maybe one restart will be required. It should only be changed
if computational resources demand it. However, the treatment of core ionization or core excitation processes
often requires a large value of NDav. At restart, Ritz vectors are copied as new trial vectors for all roots,
which will then be orthonormalized, while new vectors will only be generated for the non-converged roots.
This means that the step after the rebuilding of the expansion space will be 1-2 times as expensive as one of
the initial steps.
New directions (trial vectors) are generated from the preconditioned residual vectors. If no preconditioning is
applied (the preconditioner is taken to be a unit matrix), one falls back to the Lanczos algorithm, which is
inferior to the Davidson algorithm. This happens if the keyword DoLanczos is true. This is not recommended,
as the Lanczos algorithm converges several times slower than Davidsons, and is there for debugging mainly.
The original Davidson preconditioner is the inverse of a diagonal matrix which contains the difference of the
diagonal elements of the Hamiltonian and the current approximation to the eigenvalue belonging to the given
root. If Ria and Rijab are elements of the singles and doubles amplitudes, respectively, then the updated
vectors (Tia , Tijab ) have the form
Ria
Tia = (9.263)
Dia + ER
for singles, and
Ria
Tijab = (9.264)
Dijab + ER
for doubles. Here, Dia and Dijab are related to, and possibly approximations of, the respective diagonal
Hamiltonian elements. The simplest approximation is just to construct these from diagonal Fock matrix
elements (i.e. orbital energies) as Dia = a i and Dijab = a + b i j . A slightly better preconditioning
can be obtained as follows. For singles, take the exact CIS diagonal elements, Dia = a i + g iiaa , where the
last term is the respective antisymmetrized integral; and construct the doubles as Dijab = Dia + Djb . This is
the default, and can be changed back to the simple Fock matrix guess by setting UseCISUpdate to false.
Following the preconditioning step, the resulting vectors are orthogonalized to the previous set of trial vectors,
and orthonormalized among themselves. Since, the trial vectors do not change once they are generated (unless
a restart occurs), only the new elements of the overlap matrix need to be generated for the orthonormalization.
The numerical threshold for the inversion (and other division steps) is controlled by the parameter OTol.
Finally, the amount of printed information can be controlled via the PrintLevel keyword. If not given or
equal to 2, only basic iteration information will be printed. If set to 3, detailed iteration information will be
printed (recommended if timing results for individual steps are required), while 4 or higher triggers additional
(and very verbose) information from other subroutines as well.
The default solver in the RHF case is a multi-root Davidson procedure. The single-root solver can be initiated
setting DoRootwise true and FollowCIS to true. The latter is more stable, when large number of roots are
requested. The single-root solver is also currently the default in the UHF EOM-CCSD implementation, since
the multiroot solver can be unstable (its use is not recommended at present).
9.20.7 Properties in the RHF EOM implementation
The only property that can be calculated with the current RHF EOM implementation is the transition
moment. It is calculated as a CI-like expectation value, as proposed by Stanton and Bartlett. The right and
left transition density are defined as
GrEx
pq = h0 |(1 + )[eT {p+ q }eT , R]|0 i (9.265)
ExGr
pq = h0 |LeT {p+ q }eT |0 i (9.266)
In the above equation, corresponds to the ground state left vector, which needs to solved once and L is the
left vector , which needs to be solved separately for each root. Once the right and left vectors have been
obtained, the left and right transition densities are constructed and the oscillator strength is calculated using
following formula
2
f = |pq ExGr
pq
GrEx
||pq pq | (9.267)
3
The oscillator strength, calculated by default, employs a linear approximation for . The L vectors are, on
the other hand, calculated as a general inverse of the corresponding R vectors. This approximation requires
no additional effort over the energy calculation and gives similar accuracy as that of the exact oscillator
strength calculation, which is at least twice the cost of the energy calculation. Exact EOMCC transition
moments can, however, be calculated by setting DoLeft and DoTDM to true. Please note that transition
moments have not yet been implemented for the UHF EOM-CCSD approach.
9.20.8 Some tips and tricks for EOM-CC calculation
The COSX approximation gives significant savings in terms of memory use, disk space use and
computational timings without almost no loss of accuracy [383]. Therefore, the preferred setting for
large scale calculations should include DoCOSXEOM true,DoAOX3e true and KCOpt KC AOBLAS
(N.B. COSX is only possible in the RHF EOM implementation, in the UHF EOM implementation
AOX is currently the only option that can be used) .
The EOMCC code in ORCA has three version of the Davidsons solver. In the closed-shell case, the
default one is multi-root solver which does optimization of all the roots together. It gives the fastest
convergence and is more suitable when one is interested only in a few roots of a big molecule. However,
the multi-root solver can land into numerical issues, if more than 10 root are desired. In that case, one
can invoke the root-wise solver by setting DoRootwise true. The single root solver is very stable and
should be used when large number of roots are desired (i.e. the single root solver is also the default
in the UHF implementation, since the multi-root solver can be unstable). However, the convergence
of the single root solver is slower than the multi-root one. In the RHF implementation, there is also
a batchwise solver, where a subset of the total number of roots is optimized together. This can be
invoked by setting NRootsPerBatch to true and is intermediate between the multi-root and single-root
solver in terms of stability and convergence.
If the EOM iterations do not converge within 50 cycles, one can try to increase the number of iterations
by setting MaxIter to a larger value. One can also try to increase the dimension of the Davidsons
space by increasing the NDAV value and this generally helps in convergence acceleration. However,
setting NDAV to a value larger than 200 can make the calculation prohibitively costly .
Convergence thresholds of DTol 1e-5 (Davidson convergence) and STOL 1e-7 (ground state CCSD
convergence ) generally yield sufficiently converged energies, and are suitable for most purposes.
The normal Davidson solver generally leads to the lowest energy solutions. This procedure can also
yield roots dominated by double excitations (the so-called satellite states) for the IP and EA variants
of EOMCC, when one asks for a large number of roots. If one interested in the low lying Koopmans
type of IP and EA states, they can be obtained by setting FOLLOWCIS to true. This will follow the
initial guess provided by the Fock operators.
9.21 Excited States via STEOM-CCSD
The EOM-CCSD approach for excitation energies becomes prohibitively costly for large systems because of its
O(N 6 ) scaling. Therefore, one needs a more compact form of the wave-function ansatz. A second similarity
transformation can compress the final matrix diagonalization step to the CIS space only. The resulting
STEOM-CCSD method of Marcel Nooijen and co-workers is an efficient way for accurate calculations of
excitation energies
9.21 Excited States via STEOM-CCSD 553
In the standard EOMCC method, the transformed Hamiltonian is diagonalized over a singles and doubles
space to obtain ionized, attached, or excited states of the reference state. In STEOMCC, one performs a
second similarity transformation
G S }
= {eS }1 H{e (9.268)
including singles and doubles, is defined as

The transformation operator S,
S = SIP + SEA (9.269)
im0 + 1 Sij
SIP = Sim0 E mb mb
Eij (9.270)
2
0
SEA = Sea E a0 + 1 S ab E
ab (9.271)
e
2 ej ej
In the above equations, m and e denote active indices of the hole and particle type respectively, while a
prime denotes a restriction to orbitals that are not active. The amplitudes of the operator S are defined
in such a way that matrix elements of the transformed Hamiltonian, in second quantized notation, become
equal to zero.
0
gim0 = gij
mb
= gea = gej
ab
=0 (9.272)
after solving the CCSD equations, remain preserved. The

In addition, the zeros which pre-existed in H,

above equations are linear in S and are equivalent to the Fock space multireference coupled cluster equations
for the one valence problem. However, to ensure numerical stability, the equations are re-casted as matrix
diagonalization problem and solved as EOMIP-CCSD and EOMEA-CCSD problems. The SIP and SEA
are extracted from converged EOMIP-CCSD and EOMEA-CCSD calculations, respectively, by invoking
intermediate normalization on the suitably chosen eigenvectors corresponding to active holes and active
particles. The total process can be described as following
Solution of the ground state coupled cluster equations
= eT He
Construct the first similarity transformed Hamiltonian as H T
Solution of the EOMIP and EOMEA equation
Extraction of the S amplitudes

= eS He
Construct the second similarity transformed Hamiltonian as G S
in CIS space

Diagonalization of G
The advantage of the above method is that, instead of one iterative O(N 6 ) scaling diagonalization step, it
requires two iterative O(N 5 ) scaling steps, one non-iterative O(N 5 ) scaling step and one iterative O(N 4 )
scaling matrix diagonalization step. The presence of so-called disconnected triples terms ensures the
charge transfer separability of the excited states, which is absent even in EOM-CCSD. In addition, since
the final diagonalization step is performed in a CIS space, the spin adaption is trivial and excited states
of triplet multiplicity can be obtained without going through the complications of a spin orbital based
implementation.
All the speed up options, including CCSD(2) and COSX, which are available for EOM-CCSD are also available
for STEOM. The most important step in a STEOMCC calculation is the EOMIP-CC and EOMEA-CC
calculation steps. These steps are performed using the EOM-CCSD module and the relevant keywords are
the same as that described in 9.20. The keywords which are exclusive to the STEOM module are:
%mdci
#STEOM parameters - defaults displayed
DoCISNAT true # automatic selection of active space
NACTIP 3 # number of states defined as active in the IP calculation
NACTEA 2 # number of states defined as active in the EA calculation
DOTRIPLET false # target state of triplet multiplicity
Dodbfilter true # filters out states with doubles excitation character
DONEWACTSCH true # new active space selection scheme for STEOM-CCSD
DoSOLV # perturbative correction for solvation effects (experimental)
#Default values for automatic active space selection scheme
OTHRESH 0.001 # Cut of occupation of CIS natural orbitals in the IP calculation
VTHRESH 0.001 # Cut off occupation of CIS natural orbitals in the EA calculation
IPSTHRS 80 # The percentage singles threshold for the IP calculation
EASTHRS 80 # The percentage singles threshold for the EA calculation
end
9.21.2 Selection of Active space
The results of a STEOMCC calculation depend upon the number of roots selected as active in the EOMIP
and EOMEA calculations. In ORCA, they are chosen automatically, by using state-averaged CIS natural
transition orbitals (NTO). The orbitals up to a predefined occupation are chosen to be active in the EOMIP
and EOMEA calculations, and this is controlled by the keywords OTHRESH and VTHRESH respectively.
Now, there is two possible ways to chose active space. One is to use the criteria of percentage occupation of
NTOs as described in ref [384]. However, a newer and more robust approach is to use the criteria of absolute
occupation, which is default in the current implementation. One can switch on the old percentage occupation
based active space selection by setting DONEWATSCH to true.
One can also select the active spaces manually by turning the DoCISNAT to false, and setting the NACTIP
and NACTEA to desired values. However, this is not recommended for general users. The following shows
the output of the active orbital selection procedure:
------------------------------------------
STATE AVERAGED NATURAL ORBITALS FOR ACTIVE SPACE SELECTION
------------------------------------------
Solving eigenvalue problem for the occupied space ... Occupied block occupation :
0 0.000478
1 0.002266
2 0.169928
3 0.171663
4 0.310125
5 0.345541
Orbital taken as active for IP roots:
0 0.345541
1 0.310125
2 0.171663
3 0.169928
done
Solving eigenvalue problem for the virtual space ... Virtual block occupation :
6 0.640886
7 0.332262
8 0.017272
9 0.005326
10 0.001752
11 0.000667
12 0.000574
13 0.000540
14 0.000160
15 0.000150
16 0.000139
17 0.000086
18 0.000082
19 0.000037
20 0.000023
21 0.000016
22 0.000013
23 0.000008
24 0.000003
25 0.000002
26 0.000001
27 0.000000
28 0.000000
29 0.000000
30 0.000000
31 0.000000
32 0.000000
33 0.000000
34 0.000000
35 -0.000000
Orbital taken as active for EA roots :
0 0.640886
1 0.332262
2 0.017272
done
No of roots active in IP calculation: 4
No of roots active in EA calculation: 3
9.21.3 The reliabilty of the calculated excitation energy
The excitation energy for any states calculated in STEOMCC are only reliable when the dominant excitation
for that states are confined within the active space. This can be verified from the percentage active character
of the calculated states, a posteriori diagnostic which is defined as

P
(m, e) C(m, e)
m,e
%active character = P 100 (9.273)
(i, a) C(i, a)
i,a
where, m and e are the no active occupied and virtual orbital. The roots which has %active character higher
than 98.0 are considered to be converged with respect to the active space.
------------------
STEOM-CCSD RESULTS
------------------
IROOT= 1: 0.145412 au 3.957 eV 31914.3 cm**-1

-0.169361 4 -> 8
-0.984822 7 -> 8
Percentage Active Character 99.86

-0.166580 4 -> 8
-0.975432 7 -> 8
-0.124356 7 -> 13
IROOT= 2: 0.309409 au 8.419 eV 67907.5 cm**-1

0.994141 7 -> 9

-0.990029 7 -> 9
IROOT= 3: 0.336993 au 9.170 eV 73961.4 cm**-1

-0.994078 5 -> 8

-0.984116 5 -> 8
-0.136769 5 -> 13
IROOT= 4: 0.357473 au 9.727 eV 78456.2 cm**-1

0.181761 4 -> 10
0.728209 6 -> 8
0.611668 7 -> 10
-0.191540 7 -> 12
Warning:: the state may have not converged with respect to active space
-------------------- Handle with Care --------------------

-0.184144 4 -> 10
-0.725183 6 -> 8
-0.633718 7 -> 10
IROOT= 5: 0.386654 au 10.521 eV 84860.8 cm**-1

0.980406 4 -> 8
-0.178551 7 -> 8

0.971678 4 -> 8
0.122877 4 -> 13
-0.179242 7 -> 8
IROOT= 6: 0.444881 au 12.106 eV 97640.1 cm**-1

-0.995150 6 -> 9

-0.989966 6 -> 9
If the %active character for any calculated state is less than 98, that state may have not converged with
respect to active space and the excitation energy for that particular state is less reliable. The user should ask
for more no of roots under those conditions.
9.21.4 Removal of IP and EA states with double excitation character
To obtain accurate results with STEOM-CCSD, only the S amplitudes corresponding to the states dominated
by single excitations should be included in the second similarity transformation. This is ensured in ORCA
in two ways. First, the root following (FOLLOWCIS) is activated by default so that it converges to the
states dominated by singly excited guess vectors. This avoids the calculation of so called satellite states,
which are of double excitation character with respect to the ground state. Secondly, among the converged IP
and EA roots, the states which have %singles character below a certain predefined threshold (i.e. controlled
by the keywords IPTHRESH and EATHRESH) are automatically excluded from the second similarity
transformation.
EOM-CCSD RESULTS
----------------
IROOT= 1: 0.105316 au 2.866 eV 23114.2 cm**-1

0.697547 x -> 8
IROOT= 2: 0.217925 au 5.930 eV 47829.1 cm**-1
-0.701454 x -> 9
IROOT= 3: 0.304098 au 8.275 eV 66741.8 cm**-1

-0.700458 x -> 10
IROOT= 4: 0.350387 au 9.535 eV 76901.1 cm**-1
0.702705 x -> 11
IROOT= 5: 0.651462 au 17.727 eV 142979.4 cm**-1
0.637352 x -> 12
0.121747 x -> 8 4 -> 10
0.177039 x -> 8 5 -> 9
0.109987 x -> 9 5 -> 8
-0.206789 x -> 8 7 -> 10
-0.109870 x -> 10 7 -> 8
EA STATE= 0: percentage singles 95.282

Warning: high double excitation character, excluding from the STEOM transformation
Final no active EA roots: 4
Note that the use of CIS natural transition orbitals can lead to convergence issues for the IP and EA states
which are dominated by double excitation character. This can be remedied by setting Dodbfilter to true.
9.21.5 Properties
Presently, the only property which can be calculated with the current implementation is the transition
moment. It is calculated using a simple CIS-like formulation, employing the converged STEOM eigenvectors.
The transition moments are computed by default in a STEOM calculation.
9.21.6 Some tips and tricks for STEOMCC calculations
The accuracy of results depends upon the threshold defining the occupation of the NTOs which are
taken as active. The following threshold settings are optimal and can be used for almost all the
calculations.
OTHRESH 0.001
VTHRESH 0.001
end
The convergence problems with the underlying CIS for NTO generation can be removed by setting
Maxiter to a high value.
9.22 Excited States using PNO based coupled cluster 559
9.22 Excited States using PNO based coupled cluster
Despite the successes of the DLPNO-CC approximation for ground states, the use of PNOs for excited states
has been less fruitful. It is not straightforward to define a PNO based scheme for excited states, which will
maintain the balance between speed and accuracy, as observed for the ground state. As an intermediate
solution, the ground state DLPNO quantities are transformed back to the canonical basis and are used
within the canonical EOM routine. This procedure is justified, as the main bottle neck of the EOM-CCSD or
STEOM-CCSD methods comes from the ground state calculation. Approximating the ground state CCSD
amplitudes with MP2 amplitudes is also possible, as in the EOM-CCSD(2) approach. However, it is not very
reliable and can lead to very large errors when the reference Hartree-Fock wave-function does not provide a
correct zeroth order description of the ground state.
The back-transformation of the ground state DLPNO-CCSD amplitudes to the virtual space involves three
steps. The T amplitudes in the PNO basis are first converted into the PAO basis, then subsequently to the
atomic orbital (AO) basis, and finally to the canonical MO basis [385],
dij

a
ij
ij Ta
b
dij
b
Lij

ij ij
T
L
ij ij ij ij
Ca T Cb Tab , (9.274)
ij ij ij

diaii Taiii Li Ti Ca
i
Ti Tai , (9.275)
The AO basis functions are denoted as , , . . ., while

, , . . . refers to PAOs. The missing pairs are treated
using MP2 amplitudes. If all the thresholds are set to zero, the back-transformed amplitudes match exactly
with the canonical RI-EOM-CCSD ones. On the other hand, when all the thresholds are made infinitely
tight, one obtains the EOM-CCSD(2) results. This PNO based excited state approach is available for all the
flavors of EOM-CCSD and for STEOM-CCSD.
Below, we list all the parameters that influence the DLPNO-CCSD based excited state calculations
%mdci
#bt-PNO-EOM and STEOM parameters - defaults displayed
DoEOMMP2 true # MP2 correction for missing pairs
DoRECAN true # recanonicalization of the occupied
# orbitals before the excited state calculation
end
9.22.2 Reference State Energy
Here it should be noted that the reference energy for PNO based EOM-CCSD or STEOM-CCSD is slightly
different from that printed from a converged ground state DLPNO-CCSD calculation, as it includes the
perturbative correction for different truncated quantities.
----------------------
COUPLED CLUSTER ENERGY
----------------------
E(0) ... -113.876105722

E(CORR)(strong-pairs) ... -0.332761023
E(CORR)(weak-pairs) ... -0.000059148
E(CORR)(corrected) ... -0.332820172
E(TOT) ... -114.208925894
In the bt-PNO-EOM-CCSD scheme, the CI-like excited state treatment of the reference state is defined by
back transformed DLPNO amplitudes (or MP2 amplitudes for the weak pairs). The energy corresponding to
this set of amplitude is printed at the beginning of the EOM calculations.
Dressing integrals for EOM-CCSD ...

Making TAU ... done ( 0.0)
Reference state energy for bt-PNO-EOM-CCSD ... -114.208868242
Making FD ... done ( 0.7)

Making IKJLD ... done ( 0.0)
Making IKJADs,IKJAD ... done ( 0.9)
Making IJABD,IAJBD ... done ( 0.1)
Making IBACDs,IBACD ... done ( 0.1)
done ( 1.8)
Therefore, to calculate the total energy of an excited (ionized or electron attached) state, one needs to add
the excitation energy to the reference state energy in bt-PNO-EOM-CCSD.
9.22.3 Use of Local Orbitals
The use of local orbitals makes it difficult to follow a particular guess vector in the Davidson digonalization
process in EOM-CC and STEOM-CC. Therefore, it is advisable to recanonicalize the occupied orbitals after
the ground state DLPNO-CCSD calculation by setting DoRECAN to true. It should be noted that the
recanonicalization does not change the EOM-CCSD energies. However, the STEOM-CC energies are not
invariant to orbital rotations and differ slightly for local and canonical orbitals.
9.22.4 Some tips and tricks for bt-PNO calculations
The bt-PNO scheme with tightPNO settings gives results, which are within 0.01 eV of the canonical
EOM-CCSD numbers, at a fraction of the computational cost [385]. So, use of bt-PNO scheme is
always preferable over canonical calculations.
One should set DLPNOLINEAR true and NEWDOMAINS true in the mdci block input to use the
2015 fully linear scaling implementation, which is more robust than the 2013 implementation used as
default in bt-PNO scheme.
9.23 The Multireference Correlation Module 561
The transition moment in bt-PNO-EOM and bt-PNO-STEOM is only available using the linear
approximation.
9.23 The Multireference Correlation Module
A number of uncontracted multireference approaches are implemented in ORCA and reside in the orca mrci
module. All of these approaches start with a reference wavefunction that consists of multiple configurations
(orbital occupation patterns). The reference wavefunction defined in the ref subblock can be a complete
active space (CAS), restricted active space (RAS) or an arbitrary list of configurations. The total wavefunction
is constructed by considering single and double excitations out of the reference configurations. These excited
configurations are then used to generate configuration state functions (CSF) that have the proper spin and
spatial symmetry. The number of wavefunction parameters rapidly grows with the number of reference
functions. The orca mrci module features a set of truncation criteria (TSel, TPre, TNat) that help to reduce
the number of wavefunction parameters. Furthermore, by default, the program only considers reference
configurations that already have the target spin and spatial symmetry. There are situations, where this
is undesired and the restrictions can be lifted with the keyword rejectinvalidrefs false. For more
information on the theory, the program module as well as its usage we recommend the review article by
Neese et al. [386]. A tutorial type introduction to the subject is presented in chapter 9.23 of the manual
and more examples in the CASSCF tutorial. The detailed documentation of all features of the MR-CI and
MR-PT module is somewhat premature and at this point only a summary of keywords is given below. A
thorough description of all technical and theoretical subtleties must wait for a later version of the manual.
The overall scaling of uncontracted approaches is steep. Hence, the methodology is restricted to small
reference spaces and small molecules in general. Note that all integral must be kept in memory!
Internally contracted multireference approaches such as NEVPT2 do not share these bottlenecks. Aside from
NEVPT2, ORCA features a fully internally contracted MRCI (FIC-MRCI) that resides in the orca autoci
module. For more details on the FIC-MRCI we refer to section 9.16.
%mrci
# -----------------------------------------------------------
# Orbital selection
# NOTE: The orbitals are used as supplied. Thus, the ORDER of
# orbitals is critical. Say you have
# nact electrons in the active space
# nint electrons in the internal space
# nfrozen electrons
# * The first nfrozen/2 orbitals will not be included in the CI
# * The next nint/2 orbitals will be doubly occupied in all
# references
# * the nact electrons are distributed over the,say, mact
# orbitals according to the active space definitions.
# The remaining orbitals are external.
# IT IS YOUR RESPONSIBILITY THAT THE ORBITAL ORDERING MAKES
# SENSE!
# A sensible two-step procedure is:
# * generate some orbitals and LOOK AT THEM. Decide which ones
# to include in the CI.
# * re-read these orbitals with ! MORead NoIter. Perhaps use
# the "rotate" feature to reorder the MOs
# Then jump right into the CI which is defined in this se-
# cond job
#
# NOTE: the MRCI module respects the %method FrozenCore settings
# -----------------------------------------------------------
Loc 0,0,0
# Localize orbitals in the internal (first flag), active
# (second flag) and external space (third flag).
UseIVOs false
# Use improved virtual orbitals in the CI
# ---------------------------------
# Method selection
# ---------------------------------
CIType MRCI # Multireference CI (default)
MRDDCI1 # Difference dedicated CI 1-degree of freedom
MRDDCI2 # Difference dedicated CI 2-degrees of freedom
MRDDCI3 # Difference dedicated CI 3-degrees of freedom
MRACPF # Average coupled-pair functional
MRACPF2 # Modified version of ACPF
MRACPF2a # A slightly modified version of ACPF-2a
MRAQCC # Average quadratic coupled-cluster
MRCEPA_R # Multireference CEPA due to Ruttink
MRCEPA_0 # CEPA-0 approximation
SORCI # Spectroscopy oriented CI
SORCP # Spectroscopy oriented couplet pair approx.
MRMP2 # Multireference Moeller-Plesset at second order
MRMP3 # Multireference Moeller-Plesset at third order
MRMP4 # Multireference Moeller-Plesset at fourth order
# but keeping only singles and doubles relative to
# the reference configurations.
# ---------------------------------
# Selection thresholds
# ---------------------------------
Tsel 1e-6 # Selection threshold for inclusion in the CI based
# 2nd order MP perturbation theory <0|H|I>/DE(MP)
Tpre 1e-4 # Selection of configurations in the reference space
# after the initial diagonalization of the reference
# space only configurations with a weight large>Tpre
# to any root are included
AllSingles false
# include ALL SINGLES in the CI. Default is now TRUE!!!
# perturbative estimate of the effect of the rejected configurations

EunselOpt 0 # no correction
1 # based on the overlap with the 0th order wavefunction
2 # calculation with the relaxed reference space
# coefficients. This is the most accurate and only
# slightly more expensive
# For CIType=MRCI,MRDDCI and SORCI the approximate correction for

# higher excitations
DavidsonOpt Davidson1 # default
Davidson2 # modified version
Siegbahn # Siegbahns approximation
Pople # Poples approximation
# For MRACPF,MRACPF2,MRAQCC and SORCP

NelCorr 0
# Number of electrons used for computing the average coupled-
# pair correction.
# =0 : set equal to ALL electrons in the CI
# =-1: set equal to all ACTIVE SPACE electrons
# =-2: set equal to ACTIVE SPACE electrons IF inactive doubles
# are excluded (as in MRDDCI)
# >0 : set equal to user defined input value
LinearResponse false
# Use ground state correlation energy to compute the shift for
# higher roots (not recommended)
# ---------------------------------
# Natural Orbital Iterations
# ---------------------------------
NatOrbIters 0 # default
# number of average natural orbital iterations
Tnat 1e-4
# cutoff of natural orbitals. NOs with an occupation number less
# then Tnat will not be included in the next iteration
# Also, orbitals with occupation number closer than Tnat to 2.0
# will be frozen in the next iteration
Tnat2 -1
# if chosen >0 then Tnat2 is the threshold for freezing the
# almost doubly occupied orbitals. Otherwise it is set equal
# to Tnat
# ----------------------------------
# Additional flags and algorithmic
# details
# ----------------------------------
PrintLevel 2 # default. Values between 1 and 4 are possible
DoDDCIMP2 false
# for DDCI calculations: if set to true the program computes
# a MP2 like correction for the effect of inactive double
# excitations which are not explicitly included in the CI. This
# is necessary if you compare molecules at different geometries
# or compute potential energy surfaces.
# ----------------------------------
# The SORCP model
# ----------------------------------
CIType_in # First step CIType
CIType_fi # Second step CIType
Exc_in # First step excitation scheme
Exc_fi # Second step excitation scheme
Tsel_in # First step Tsel
Tsel_fi # Second step Tsel
Tpre_in # First step Tsel

Tpre_fi # Second step Tpre
# Thus, the SORCI model corresponds to CIType=SORCP with

# CIType_in MRCI CIType_fi MRCI
# Exc_in DDCI2 Cexc_fi DDCI3
# Tsel_in 1e-5 Tsel_fi 1e-5
# Tpre_in 1e-2 Tpre_fi 1e-2
# ----------------------------------
# Multirerence perturbation theory
# ----------------------------------
MRPT_b 0.02 # Intruder state avoidance PT after Hirao (default 0.0)
# with this flag individual intruders are shifted away to
# to some extent from the reference space
MRPT_shift 0.3 # Level shift introduced by Roos which shifts the entire
# excited manifold away in order to avoid intruder states.
# A correction is applied afterwards but results do depend
# on this (arbitrary) value to some extent.
H0Opt projected # use an off-diagonal definition of H0
Diagonal # use a diagonal definition of H0 (much faster but maybe
# a little less reliable
Partitioning MP # Moeller plesset partitioning
EN # Epstein-Nesbet partitioning (not recommended)
Fopt Standard # Standard definition of MR Fock operators
G3 # uses Andersons g3 correction also used in CASPT2
#---------------------------------------
# restrict reference configurations
#---------------------------------------
RejectInvalidRefs true # by default reference CSFs are restricted
# to target spin and spatial symmetry
# ======================================
# Definitions of blocks of the CI Matrix
# ======================================
NewBlock 2 * # generate a Block with doublet(=2) multiplicity
Nroots 1 # number of roots to be generated
Excitations cis # CI with single excitations
cid # CI with double excitations
cisd # CI with single and double excitations
ddci1 # DDCI list with one degree of freedom
ddci2 # DDCI list with two degrees of freedom
ddci3 # DDCI list with three degrees of freedom
Flags[_class_] 0 or 1
# Turn excitation classes on or off individually
# s stands for any SOMO, i,j for internal orbitals and
# a,b for external orbitals
# Singles _class_ = ss, sa, is, ia
# Doubles _class_ = ijss, ijsa, ijab,
# isss, issa, isab,
# ssss, sssa, ssab
# Flags takes priority over Excitations. In fact Excitations
# does nothing but to set Flags. So, you can use Excitations
# to provide initial values for Flags and then modify them
# with subsequent Flags assignments

refs
#
# First choice - complete active space
#
CAS(nel,norb) # CAS-CI reference with nel electrons in
# Norb orbitals
#
# Second choice - restricted active space
#
RAS(nel: m1 h/ m2 / m3 p)
# RAS-reference with nel electrons
# m1= number orbitals in RAS-1
# h = max. number of holes in RAS-1
# m2= number of orbitals in RAS-2 (any number of
# electrons or holes)
# m3= number of orbitals in RAS-3
# p = max. number of particles in RAS-3
#
# Third choice - individually defined configurations
#
{ 2 0 1 0}
{ 1 1 1 0}
etc.
# define as many configurations as you want. Doubly occupied MOs
# singly occupied MOs and empty MOs. Important notes:
# a) the number of electrons must be the same in all references
# b) the number of orbitals is determined from the number of
# definitions. Thus, in the example above we have three active
# electrons and four active orbitals despite the fact that the
# highest orbital is not occupied in any reference.
# The program determines the internal, active and external spaces
# automatically from the number of active electrons and orbitals
end
end
# there can be as many blocks as you want!!!
# ----------------------------------
# Density matrix generation flags
# First Key= State densities <I|D|I>
# =0: none
# =1: Ground state only (lowest root of all blocks; Electron only)
# =2: Ground state only (Electron and spin density)
# =3: Lowest root from each block (Electron density)
# =4: Lowest root from each block (Electron and spin density)
# =5: All states (Electron density)
# =6: All states (Electron and spin density)
# Second Key= Transition densities <I|D|J>
# needed for all transition intensities, g-tensor etc
# =0: none
# =1: from the ground state into all excited states (el)
# =2: from the ground state into all excited states (el+spin)
# =3: from all lowest states into all excited states (el)
# =4: from all lowest states into all excited states (el+spin)
# =5: all state pairs (el)
# =6: all state pairs (el+spin)
# Note that for perturbation theory the density is computed as

# an expectation value over the first (second) order wavefunction.
# which is renormalized for this purpose
# ----------------------------------
Densities 1,1
# ----------------------------------
# Complete printing of the wavefunction
# ----------------------------------
PrintWF 1 # CFG based printing (default)
det # Determinant based wavefunction printing
TPrintWF 3e-3 # Threshold for the printing of the CFGs/Dets
# ----------------------------------
# Algorithm for the solver
# ----------------------------------
Solver Diag # Davidson like solver
DIIS # DIIS like solver
# both solvers have their pros and cons. The DIIS may converge
# better or use less time since it only recomputes the vectors that
# have not yet converged; The DIIS may be less sensitive to root flipping
# effects but occasionally it converges poorly and states of the same
# symmetry are occasionally a little problematic
# For perturbation theory DIIS is always used.
# For both solvers
MaxIter 100 # the maximum number of iterations
Etol 1e-6 # convergence tolerance for energies in Eh
Rtol 1e-6 # convergence tolerance for residual
# For Solver=Diag (Davidson solver)

Otol 1e-16 # Orthogonality threshold for Schmidt process
NGuessMat 512 # Dimension of the guess matrix 512x512
# be used to compute the initial guess of the actual MRCI calculation
NGuessMatRefCI 512 # Dimension of the guess matrix
# for the reference CI

# For the Solver=DIIS. Particularly recommended for anything else but
# straightforward CI and also for calculations in direct2 mode!
MaxDIIS 5 # Maximum number of old vectors to be used in DIIS
RelaxRefs true # Relax reference space coefficients in the CI or
# freeze them to their zeroth order values
LevelShift 0.4 # Level Shift for stabilizing the DIIS procedure
# ----------------------------------
# RI Approximation
# ----------------------------------
IntMode RITrafo #Use RI integrals
FullTrafo #No RI (default)
# ----------------------------------
# Integral storage, memory and files
# ----------------------------------
IntStorage FloatVals
DoubleVals (default)
# store integrals with float (4 byte) or double (8 byte)
# accuracy in main memory

FourIndexInts false (default)
True
# Store ALL four index integrals over Mos in main memory
# only possible for relatively small systems, perhaps up
# to 150-200 MOs included in the CI
MaxMemInt 256
# Maximum amount of core memory devoted to the storage of
# integrals. If NOT all three index integrals fit into main
# memory the program fails
MaxMemVec 16
# Maximum amount of memory used for section of the trial and
# sigma vectors. This is not a particularly critical variable
KeepFiles false
# Keep integrals and CI program input file (.mrciinp). Then
# you can manually edit the .mrciinp file which is a standard
# ASCII file and run the MRCI program directly. The only thing
# you cannot change is the orbital window.
end
9.23.2 Properties Calculation Using the SOC Submodule
9.23.2.1 Zero-Field Splitting
The spin-orbit coupling (SOC) and spin-spin coupling (SSC) contributions to the zero-field splitting (ZFS)
can be calculated very accurately using a wavefunction obtained from a multiconfigurational calculation of a
multi-reference type such as CASSCF, MRCI, or MRPT in the form of:
SS X
Cl SS

I = (9.276)

Here the upper indices SS stand for a wave function of the spin quantum number S and spin projection

MS = S. Since the BO Hamiltonian does not contain any complex-valued operator, the solutions SS
I may
be chosen to be real-valued.
The SOC and SSC effects along with the Zeeman interaction can be included by means of the quasi-degenerate
perturbation theory (QDPT). In this approach the SOC, the SSC, and the Zeeman operators are calculated

in the basis of pre-selected solutions of the BO Hamiltonian SM
I .
D 0 0E D 0 0E
(S) SM
SM
SM SM
I HBO + H SOC + H SSC + H Z J = IJ SS 0 M M 0 E
I + I H SOC + H SSC + H Z J
(9.277)
Diagonalization of this matrix yields the energy levels and eigenvectors of the coupled states. These
BO with complex coefficients.
eigenvectors in fact represent linear combinations of the solutions of H
The effective one-electron SOC operator in second quantized form can be written as:
SOMF = 1
X h i
+
H zpq a
p bq + zpq bp a 0
q + zpq a q bpbq
p a (9.278)
2 pq
Here ap and bp stand for creation of and electrons respectively; a

p and bp represent the corresponding
x y + x y 0 z
annihilation operators. The matrix elements zpq = zpq izpq , zpq = zpq + izpq , and zpq = zpq (upper x, y,
z indices denote the Cartesian components) are constructed from the matrix elements described in section
9.28.3.3.
The SSC Hamiltonian reads:
2 2 m (2)
SSC = 3ge (1)
X X (2)
H 5 [rij rij ]m [S (i) S (j)] (9.279)
8 m=0,1,2
rij
i6=j m
For matrix elements between states of the same multiplicity it can be simplified to
D E
0 0 (S+1)(2S+3)
aSM H SSC a SM =
S(2S1)
(9.280)
!
0
m S 2 S P (m)

aSS Q0pqrs a0 SS
P
(1) 0
pqrs Dpqrs
m M m M
Here

1 1 z z
Q(0) z z

pqrs = Epq sr Sps Srq + Spq Srs Epq Ers (9.281)
4 6 2
represents the two-electron quintet density. The operators Epq = a q + bpbq and Spq
p a z
q bpbq symbolize
p a
=a
here the one-electron density operator and the spin density operator accordingly. The spatial part
3r1z r2z r1 r2
ZZ
(0) 1
Dpqrs = p (r1 ) r (r2 ) 5 q (r1 ) s (r2 ) dr1 dr2 (9.282)
6 r12
denotes the two-electron field gradient integrals. These two-electron integrals can be evaluated using the RI
approximation.
Finally, the Zeeman Hamiltonian is included in the form of:

Z = B L
H + ge S
B (9.283)
representing the total orbital momentum operator, and S

with L being the total spin operator.
Since the both the energies and the wavefunction of the low-lying spin-orbit states are available, the effective
Hamiltonian theory can be used to extract EPR parameters such as the full G and/or ZFS tensors. In this
approach one starts by constructing a model Hamiltonian, which for the mononuclear complexes is:
mod = SD
H S (9.284)
provided that the ground state is non-degenerate. By applying this Hamiltonian on the basis of the model
space, i.e. the |S, MS i components of the ground state, the interaction matrix is constructed.
The construction of effective Hamiltonian relies on the information contained in both the energies and the
wavefunctions of the low-lying spin-orbit states. Following des Cloizeaux formalism, the effective Hamiltonian
reproduces the energy levels of the exact Hamiltonian Ek and the wavefunctions of the low-lying states
projected onto the model space :
eff |
H k i = Ek |
ki (9.285)
These projected vectors are then symmetrically orthonormalized resulting in an Hermitian effective Hamilto-
nian, which can be written as:
1
k iEk hS 12
X
eff |i
H = |S 2 k| (9.286)
k
The effective interaction matrix obtained by expanding this Hamiltonian into the basis of determinants
belonging to the model space, is the compared to the matrix resulted from expanding the model Hamiltonian.
Based on a singular value decomposition procedure, all 9 elements of the G and/or ZFS tensors are extracted.
In the input file the relevant keys are located in the soc sub-block:
%mrci
soc
DoSOC true # include the SOC contribution
DoSSC true # include the SSC contribution
PrintLevel 2 # printing level
TPrint 0.01 # threshold for printing eigenvalue contributions
end
end
Firstly, the SSC contribution to ZFS is calculated. The output file for a CASCI calculation on top of a
CASSCF(8,6) reference of an oxygen molecule with def2-SVP basis set and four singlets and four triplet
states included into the QDPT treatment is presented as follows:
---------------------
SSC MATRIX GENERATION
---------------------
Blocks order: 1 0
Ground state mult 3

Ground state block 1
Ground state root 0
Calculated reduced SSC matrix elements

BLOCKI(Mult) BLOCKJ(Mult) Root I Root J dxx dyy dzz dxy dxz dyz
1/cm 1/cm 1/cm 1/cm 1/cm 1/cm
1( 3) 1( 3) 0 0 -0.316899 -0.316899 0.633797 -0.000000 -0.000000 -0.000000
1( 3) 1( 3) 1 0 0.000000 0.000000 -0.000000 -0.000000 -0.000000 -0.000000
1( 3) 1( 3) 1 1 0.251010 0.251010 -0.502020 0.000000 -0.000000 -0.000000
1( 3) 1( 3) 2 0 -0.000000 -0.000000 0.000000 0.000000 -0.000000 -0.000000
1( 3) 1( 3) 2 1 -0.000000 0.000000 0.000000 0.000000 -0.000000 -0.000000
1( 3) 1( 3) 2 2 0.251010 0.251010 -0.502020 0.000000 -0.000000 -0.000000
1( 3) 1( 3) 3 0 -0.000000 0.000000 -0.000000 0.000000 -0.000000 -0.000000
1( 3) 1( 3) 3 1 -0.095105 0.095105 -0.000000 -0.010169 -0.000000 -0.000000
1( 3) 1( 3) 3 2 -0.010169 0.010169 -0.000000 0.095105 -0.000000 -0.000000
1( 3) 1( 3) 3 3 0.307060 0.307060 -0.614121 -0.000000 -0.000000 -0.000000
Next, the SSC matrix is constructed form the precalculated matrix elements and subjected to diagonalization.
The full SSC matrix is available with the PrintLevel key set to 3. Then, the SSC energy levels are printed
in cm1 and eV along with the Boltzmann level populations.
Energy levels (cm-1,eV) : Boltzmann populations for T = : 300.000 K

0 : 0.000 0.0000 3.34e-01
1 : 0.951 0.0001 3.33e-01
2 : 0.951 0.0001 3.33e-01
3 : 7130.461 0.8841 4.71e-16
4 : 7130.996 0.8841 4.69e-16
5 : 13306.276 1.6498 6.45e-29
6 : 46916.451 5.8169 6.38e-99
7 : 48790.859 6.0493 7.95e-103
8 : 48790.859 6.0493 7.95e-103
9 : 48790.859 6.0493 7.95e-103
10 : 48790.859 6.0493 7.95e-103
11 : 48791.612 6.0494 7.93e-103
12 : 48791.612 6.0494 7.93e-103
13 : 49948.158 6.1928 3.09e-105
14 : 49948.158 6.1928 3.09e-105
15 : 49949.079 6.1929 3.08e-105
Additionally, the eigenvectors of the SSC matrix are printed if the PrintLevel key is set to 2:
Eigenvectors :
The threshold for printing is 0.0100000
Weight Real Image : Block Root Spin Ms
STATE 0 : 0.00
1.000000 0.999398 -0.034683 : 1 0 1 0
STATE 1 : 0.95
0.490154 -0.700086 -0.005795 : 1 0 1 1

0.509846 0.714011 0.005862 : 1 0 1 -1
STATE 2 : 0.95
0.509846 0.714010 -0.005997 : 1 0 1 1

0.490154 0.700085 -0.005927 : 1 0 1 -1
STATE 3 : 7130.24
1.000000 1.000000 -0.000000 : 0 0 0 0
STATE 4 : 7130.77
1.000000 1.000000 -0.000000 : 0 1 0 0
The SSC coupled states are counted here from the lowest one, and the energy of a state is printed in cm1 .
The Weight column represents here squared absolute values of the complex contribution coefficients of
the BO Hamiltonian states to the SSC coupled states. Real and Image are real and imaginary parts of
complex coefficients. Accordingly, Root, Spin, and Ms denote here the spin quantum number and
the spin projection number of a particular contribution. The column Block represent the symmetry block
number. Note that the program re-aranges blocks from the oder given in the input file.
Next, the SSC contribution to the ZFS obtained with first-order perturbation theory is printed along with
the D values and the E/D ratio.
---------------------------------
ZERO-FIELD SPLITTING
(SPIN-SPIN COUPLING CONTRIBUTION)
---------------------------------
raw-matrix (cm**-1):
-0.316896 -0.000000 -0.000000
-0.000000 -0.316896 -0.000000
-0.000000 -0.000000 0.633791
diagonalized D-tensor (cm**-1):
-0.316896 -0.316896 0.633791
0.000000 1.000000 0.000000

1.000000 0.000000 0.000000
0.000000 0.000000 1.000000
Direction X=1 Y=0 Z=2

D = 0.950687 cm**-1
E/D = 0.000000
Here the eigenvalues are printed in the increasing order, and the column under the eigenvalue represents the
eigenvector belonging to this eigenvalue. The axes for the coordinate system that diagonalizes the D tensor
are chosen to fulfill the condition
1
0 6 E /D 6 (9.287)
3
Next, the SOC part to the splitting is calculated. Firstly, the diagonal (with respect to the spin) matrix
element type of hI | zxpq Spq
z
|J i, hI | zypq Spq
z
|J i, hI | zzpq Spq
z
P P P
|J i are evaluated between states
pq pq pq
zxpq Spq
+
zypq Spq
+
zzpq Spq
+
P P P
of the same multiplicity and hI | |J i, hI | |J i, hI | |J i between states of
pq pq pq
different multiplicities.
Calculated reduced SOC matrix elements
BLOCKI(Mult) BLOCKJ(Mult) Root I Root J <I|Lx S |J> <I|Ly S |J> <I|Lz S |J>
1/cm 1/cm 1/cm
0( 3) 0( 3) 0 0 -0.00 0.00 -0.00

0( 3) 0( 3) 1 0 -0.00 0.00 -0.00
0( 3) 0( 3) 1 1 -0.00 -0.00 -0.00
0( 3) 0( 3) 2 0 -0.00 -0.00 -0.00
0( 3) 0( 3) 2 1 -0.00 0.00 -289.98
0( 3) 0( 3) 2 2 -0.00 -0.00 -0.00
0( 3) 0( 3) 3 0 0.00 -0.00 -0.00
0( 3) 0( 3) 3 1 0.00 -0.00 0.00
0( 3) 0( 3) 3 2 0.00 -0.00 0.00
0( 3) 0( 3) 3 3 -0.00 -0.00 -0.00
0( 3) 1( 1) 0 0 -0.00 0.00 -0.00
0( 3) 1( 1) 0 1 -0.00 0.00 -0.00
0( 3) 1( 1) 0 2 -0.00 0.00 -225.16
0( 3) 1( 1) 0 3 0.00 0.00 0.00
0( 3) 1( 1) 1 0 -0.00 0.00 0.00
0( 3) 1( 1) 1 1 0.00 0.00 -0.00
0( 3) 1( 1) 1 2 0.00 0.00 0.00
0( 3) 1( 1) 1 3 -0.00 0.00 -0.00
0( 3) 1( 1) 2 0 0.00 0.00 -0.00
0( 3) 1( 1) 2 1 0.00 -0.00 -0.00
0( 3) 1( 1) 2 2 -0.00 0.00 -0.00
0( 3) 1( 1) 2 3 0.00 -0.00 0.00
0( 3) 1( 1) 3 0 0.00 -0.00 0.00
0( 3) 1( 1) 3 1 -0.00 -0.00 -0.00
0( 3) 1( 1) 3 2 -0.00 -0.00 0.00
0( 3) 1( 1) 3 3 0.00 -0.00 205.22
The full SOC matrix elements can be printed if the PrintLevel key is set to 3. Similar to the SSC contribution,
the SOC energies are printed next.
Energy levels (cm-1,eV) : Boltzmann populations for T = : 300.000 K

0 : 0.000 0.0000 3.35e-01
1 : 1.905 0.0002 3.32e-01
2 : 1.905 0.0002 3.32e-01
3 : 7130.930 0.8841 4.71e-16
4 : 7131.465 0.8842 4.70e-16
5 : 13308.646 1.6501 6.39e-29
6 : 46856.467 5.8095 8.53e-99
7 : 48646.630 6.0314 1.59e-102
8 : 48646.630 6.0314 1.59e-102
9 : 48791.619 6.0494 7.95e-103
10 : 48791.619 6.0494 7.95e-103
11 : 48936.608 6.0674 3.97e-103
12 : 48936.608 6.0674 3.97e-103
13 : 49948.972 6.1929 3.09e-105
14 : 49948.972 6.1929 3.09e-105
15 : 49955.781 6.1937 2.99e-105
The eigenvectors of the SOC matrix are provided along with contributions from roots S, and Ms .
Eigenvectors :
The threshold for printing is 0.0100000
Weight Real Image : Block Root Spin Ms
STATE 0 : 0.00
0.999857 -0.000000 -0.999928 : 0 0 1 0
STATE 1 : 1.90
0.464214 -0.228955 0.641711 : 0 0 1 1

0.535786 0.154270 -0.715533 : 0 0 1 -1
STATE 2 : 1.90
0.535786 0.730198 0.050967 : 0 0 1 1

0.464214 0.680079 -0.041312 : 0 0 1 -1
STATE 3 : 7130.93
1.000000 1.000000 -0.000000 : 1 0 0 0
STATE 4 : 7131.46
1.000000 -1.000000 0.000000 : 1 1 0 0
Next, the SOC contribution to the ZFS obtained with second-order perturbation theory is printed along with
the D values and the E/D ratio.
----------------------------------
(SPIN-ORBIT COUPLING CONTRIBUTION)
----------------------------------
0.000000 -0.000000 -0.000000
-0.000000 0.000000 0.000000
-0.000000 0.000000 1.905043
0.000000 0.000000 1.905043
0.889614 0.456713 -0.000000

-0.456713 0.889614 0.000000
0.000000 0.000000 1.000000

D = 1.905043 cm**-1
E/D = 0.000000
Independent state contribution to the ZFS is printed if PrintLevel is set to 3.

Individual contributions to D tensor:
block Mult root D E
0 3 0 0.000 0.000
0 3 1 -0.000 0.000
0 3 2 -0.000 -0.000
0 3 3 -0.000 -0.000
1 1 0 0.000 0.000
1 1 1 0.000 -0.000
1 1 2 1.905 0.000
1 1 3 0.000 0.000
From this example is readily seen that the whole contribution to the SOC part of D-value comes from the
interaction between the ground state and the singlet third root.
The ZFS tensor based on the effective Hamiltonian theory is printed next.
--------------------------------------------------------
(EFFECTIVE HAMILTONIAN SPIN-ORBIT COUPLING CONTRIBUTION)
--------------------------------------------------------
-0.000000 0.000000 -0.000000
0.000000 0.000000 -0.000000
-0.000000 -0.000000 1.904771
-0.000000 0.000000 1.904771
0.756590 -0.653889 -0.000000

-0.653889 -0.756590 -0.000000
0.000000 -0.000000 1.000000

D = 1.904771 cm**-1
E/D = 0.000000
Finally, all these steps described above are repeated for the combined SOC and SSC matrix.
The diagonal matrix elements of the SOC and SSC matrix which correspond to non-relativistic energies can
be replaced by more accurate values. In the output of the program the corresponding transition energies
printed in cm1 .
# In case that you want to run QDPT-SOC calculation with manually

#adjusted diagonal energies you can copy the following part into
#the %mrci soc block
#and modify it as needed(energies are given in
#wavenumbers relative to the lowest state)
# NOTE: It is YOUR responsibility to make sur that the CAS-CI state
#that you may want to dress with these energies correlate properly
#with the energies printed here. The order of states or even the
#identity of states may change with and without inclusion of
#dynamic correlation In the case that dynamic correlation strongly
#mixes different CAS-CI states there may not even be a proper
#correlation!
#
EDiag[ 0] 0.00 # root 0 of block 0
Those transition energies can be substituted by a more accurate energies provided in the input file as follows:
%soc
dosoc true
dossc true
end
Accurate diagonal energies generally improve the accuracy of the SOC and SSC splittings.
9.23.2.2 Local Zero-Field Splitting
The submodule can also be used to calculate the local ZFS splitting parameters of atomic centers. The
method, referred to as local complete active space configuration interaction (L-CASCI), can be used to
separate into atomic contributions the SOC part of the total ZFS tensor. The rational behind it and additional
details are described in the original publication [387]; below are listed only the steps required to reproduce
the calculation for the dimer complex presented there.
1. The first step consists in obtaining the molecular orbitals that are going to be used in the configuration
interaction (CI) procedure. A good set of orbitals can be obtained from a restricted open-shell spin-averaged
Hartree-Fock (SAHF) calculation. The relevant part of the input is listed below:
! def2-tzvp keepfock
% scf
hftyp rohf
rohf_case sahf
rohf_numop 2
rohf_nel[1] 9
rohf_norb[1] 10
end
For the present Mn(II)Mn(III) dimer there are a total of 9 electrons distributed into 10 d-orbitals.
2. Next, the molecular orbitals are localized using one of the implemented localization schemes. Below is the
orca loc input used in this case:
sahf.gbw
sahf.loc
0
200 # first of the 10 d-orbitals
209 # last of the 10 d-orbitals
128
0.000001
0.75
0.65
2
3. Following this, the localized orbitals are made locally canonical by block diagonalizing the Fock matrix
using the orca blockf utility.
orca_blockf sahf.fsv sahf.loc 200 204 205 209
The first two numbers define the range of molecular orbitals localized on one center; the last two are for the
second center.
4. The recanonicalized orbitals stored in the sahf.loc file can be then used to calculate the SOC contribution
to the local ZFS of the Mn(III) center using the following MRCI input:
! zora-def2-tzvp def2-tzvp/c zora

! nomulliken noloewdin
! moread noiter allowrhf
! moread
% mrci
citype mrci
tsel 0
tpre 0
intmode ritrafo
solver diis
soc
intmode ritrafo
dosoc true
end
newblock 10 *
nroots 5
excitations none
refs
# Mn(II) Mn(III)
1 1 1 1 1 1 1 1 1 0
1 1 1 1 1 1 1 1 0 1
1 1 1 1 1 1 1 0 1 1
1 1 1 1 1 1 0 1 1 1
1 1 1 1 1 0 1 1 1 1
end
end
newblock 8 *
nroots 45
excitations none
refs
# Mn(II) Mn(III)
1 1 1 1 1 2 1 1 0 0
1 1 1 1 1 2 1 0 1 0
1 1 1 1 1 2 1 0 0 1
1 1 1 1 1 2 0 1 1 0
1 1 1 1 1 2 0 1 0 1
1 1 1 1 1 2 0 0 1 1
1 1 1 1 1 1 2 1 0 0
1 1 1 1 1 1 2 0 1 0
1 1 1 1 1 1 2 0 0 1
1 1 1 1 1 1 1 2 0 0
1 1 1 1 1 1 1 1 1 0
1 1 1 1 1 1 1 1 0 1
1 1 1 1 1 1 1 0 2 0
1 1 1 1 1 1 1 0 1 1
1 1 1 1 1 1 1 0 0 2
1 1 1 1 1 1 0 2 1 0
1 1 1 1 1 1 0 2 0 1
1 1 1 1 1 1 0 1 2 0
1 1 1 1 1 1 0 1 1 1
1 1 1 1 1 1 0 1 0 2
1 1 1 1 1 1 0 0 2 1
1 1 1 1 1 1 0 0 1 2
1 1 1 1 1 0 2 1 1 0
1 1 1 1 1 0 2 1 0 1
1 1 1 1 1 0 2 0 1 1
1 1 1 1 1 0 1 2 1 0
1 1 1 1 1 0 1 2 0 1
1 1 1 1 1 0 1 1 2 0
1 1 1 1 1 0 1 1 1 1
1 1 1 1 1 0 1 1 0 2
1 1 1 1 1 0 1 0 2 1
1 1 1 1 1 0 1 0 1 2
1 1 1 1 1 0 0 2 1 1
1 1 1 1 1 0 0 1 2 1
1 1 1 1 1 0 0 1 1 2
end
end
end
5. The three second order ZFS components printed at the end of the calculation (Second order D-tensor:
component 0, etc.) are scaled using the S value for the complex, which in this case is 4.5 (9 electrons 0.5).
In order to obtain the correct local value of the ZFS, the three matrices have to be rescaled using the S value
for Mn(III), which is to 2. Note that the three matrices have different scaling prefactors, and the dependence
on S is not the same:
1
DSOC(0) S2
1
DSOC(1) S(2S1)
1
DSOC(+1) (S+1)(2S+1)
These equations can be used to calculate the required prefactors. For example in the case of the SOC -(0) the
prefactor is equal to:
SOC(0) 4.52 SOC(0) SOC(0)
DMn(III) = 22 Ddimer = 5.0625 Ddimer
The final step is to scale the two remaining matrices using the appropriate prefactors, sum all three of them
up, diagonalize the resulting the matrix, and use its eigenvalues to calculate the D and E parameters. These
represent the local ZFS parameters of the Mn(III) center.
9.23.2.3 g-Tensor
The orca mrci program contains an option to calculate g-tensors using MRCI wavefunctions. For a system
with an odd number of electrons, the doubly degenerate eigenvalues obtained from the QDPT procedure
represent Kramers pairs and used to build the matrix elements of the total spin operator and the total
angular momentum operator from the Zeeman Hamiltonian. Denoting as a solution and as its Kramers
partner and using matrix element notations
k + ge Sk |i , k = h| L
k11 = h| L k + ge Sk |i , k = x, y, z (9.288)
12
The elements of g-matrix are obtained as:
gkz = 2ck11 , gky = 2c= k12 , gkx = 2c< k12

(9.289)
Then, the true tensor G is built from g-matrices:
G = gg T (9.290)
G is subjected further to diagonalization yielding positive eigenvalues, the square roots of which give the
principal values of g-matrix.
p p p
gxx = Gxx , gyy = Gyy , gzz = Gzz (9.291)
A typical mrci block of the input file for a g-tensor calculation should (e.g. for a S=3/2 problem) look as the
following:
%mrci ewin -4,1000

citype mrci
cimode direct2
intmode fulltrafo
solver diis
etol 1e-8
rtol 1e-8
tsel 1e-6
tpre 1e-5
soc
PrintLevel 2
GTensor true # make g-tensor calculations
NDoubGTensor 2 # number of Kramers doublets to account
# for every pair a separate
# calculation is performed
end
newblock 4 *
excitations cisd
nroots 10
refs cas(7,5) end
end
end
The result for the first Kramers pair is printed as follows:

--------------
KRAMERS PAIR 1
--------------
Matrix elements Re<1|S|1> -0.072128 0.024511 -2.998843

Matrix elements Re<1|S|2> -0.001088 0.000366 -0.002010
Matrix elements Im<1|S|2> -0.000354 -0.001037 -0.000173
Matrix elements Im<1|L|2> -0.000006 -0.000011 -0.000065
-------------------
ELECTRONIC G-MATRIX
-------------------
g-matrix:
-0.002240 0.000754 -0.005551
0.000720 0.002100 0.000477
-0.198556 0.067498 -8.251703
g-factors:
0.002220 0.002222 8.254370 iso = 2.752937
g-shifts:
-2.000100 -2.000098 6.252051 iso = 0.750618
Eigenvectors:
0.057426 0.998060 0.024055
0.998327 -0.057244 -0.008177
0.006784 -0.024484 0.999677
Here for the L and S matrix elements indices 1 and 2 are assumed to denote Kramers partners, and three
numbers in the first row stand for x, y, z contributions.
In addition the g-tensor is calculated within the Effective Hamiltonian formalism.
----------------------------------------------
ELECTRONIC G-MATRIX FROM EFFECTIVE HAMILTONIAN
----------------------------------------------
g-matrix:
1.978874 -0.000345 0.018908
-0.000345 1.977899 -0.006433
0.018879 -0.006418 2.763402
g-factors:
1.977789 1.978477 2.763909 iso = 2.240058
g-shifts:
-0.024530 -0.023843 0.761590 iso = 0.237739
Eigenvectors:
0.288884 0.957062 0.024060
0.957364 -0.288770 -0.008181
0.000882 -0.025397 0.999677
# The g-factors are square roots of the eigenvalues of gT*g

# Orientations are the eigenvectors of gT*g
Finally and only within the MRCI module the g-tensor is evaluated by using the Sum Over States formalism
[388]:
---------------------------------------------------------------------------
SUM OVER STATES CALCULATION OF THE SPIN HAMILTONIAN (for g and HFC tensors)
---------------------------------------------------------------------------
Ground state index = 0

Ground state multiplicity = 4
Ground state spin density = P[ 1]
State = 1 <0|P|I>= 2 <0|Q|I>= 19
State = 2 <0|P|I>= 3 <0|Q|I>= 27
State = 3 <0|P|I>= 4 <0|Q|I>= 34
State = 4 <0|P|I>= 5 <0|Q|I>= 40
State = 5 <0|P|I>= 6 <0|Q|I>= 45
State = 6 <0|P|I>= 7 <0|Q|I>= 49
State = 7 <0|P|I>= 8 <0|Q|I>= 52
State = 8 <0|P|I>= 9 <0|Q|I>= 54
State = 9 <0|P|I>= 10 <0|Q|I>= 55
Origin for angular momentum ... ( -0.0006, -0.0010, 0.0021)

Kinetic Energy ... done
Relativistic mass correction ... done
Gauge correction ... done
Angular momentum integrals ... done
Reading Spin-Orbit Integrals ... done
-----------------------
MATRIX ELEMENT PRINTING
-----------------------
Energy differences (DE=EI-E0) and spin-orbit matrix elements (SO=<I|HSO|0>) are

printed in cm**-1. Orbital Zeeman matrix elements (L=<I|L|0>) are printed in au.
State DE LX LY LZ SOX SOY SOZ
1 1349.3 0.0464 -0.0158 1.9264 -23.432 7.965 -974.312

2 13026.2 -0.6596 0.6888 0.0214 337.028 -351.116 -10.966
3 13615.1 -0.6961 -0.6514 0.0113 354.225 332.219 -5.736
4 56686.3 -0.0053 0.0077 0.0971 1.794 -1.696 -36.786
5 56954.2 -0.0516 -0.0048 -0.0042 28.211 5.821 1.459
6 56994.0 -0.0418 0.0233 -0.0025 15.185 -2.144 1.145
7 63371.5 -0.0211 0.0226 0.0078 3.833 -2.948 -2.724
8 64176.0 -0.0652 0.0032 -0.0002 32.779 6.146 0.063
9 74309.9 -0.0007 0.0032 0.0380 0.183 -1.058 -13.517
-------------------
ELECTRONIC G-MATRIX
-------------------
raw-matrix :
2.025533 -0.000738 0.021755

-0.000738 2.024537 -0.007389
0.021755 -0.007389 2.928943
g-factors:
2.024122 2.025363 2.929527 iso = 2.326338
g-shifts:
0.021803 0.023044 0.927208 iso = 0.324018
Eigenvectors:
0.533896 -0.845208 0.024064
0.845530 0.533866 -0.008182
-0.005932 0.024715 0.999677
Euler angles w.r.t. molecular frame (degrees):

-76.5038 1.4564 -161.2223
-----------------------------
CONTRIBUTIONS TO THE G-MATRIX
-----------------------------
Term g1 g2 g3
--------------------------------------------------------------------
Relativistic mass correction: -0.0008220 -0.0008220 -0.0008220
Gauge correction : 0.0000000 0.0000000 0.0000000
g(OZ/SOC) : 0.0226250 0.0238662 0.9280297
State 1 : 0.0000000 -0.0000000 0.9279829

State 2 : 0.0013767 0.0223913 0.0000000
State 3 : 0.0212332 0.0014408 0.0000000
State 4 : 0.0000000 0.0000004 0.0000418
State 5 : 0.0000074 0.0000099 0.0000001
State 6 : 0.0000002 0.0000078 0.0000001
State 7 : 0.0000000 0.0000015 0.0000002
State 8 : 0.0000076 0.0000144 0.0000000
State 9 : 0.0000000 0.0000000 0.0000046
-----------------------------------------
Total g-shifts : 0.0218030 0.0230442 0.9272077
# The g-factors are square roots of the eigenvalues of gT*g

# Orientations are the eigenvectors of gT*g
Note that within the SOS formalism in addition to the second order (SOC) contributions the bilinear to
the field terms: Relativistic mass correction and diamagnetic spin-orbit term (Gauge) are evaluated. As
can be seen these corrections are rather negligible in comparison to the second order SOC contributions
and most of the time can be safely omitted. Moreover further insight is obtained by printing the individual
contribution of each excited state to the g-tensor. In the example above the first excited state contributes to
the gz component while the next two to both the gx and gy components, respectively.
So to summarize the g-tensor calculations in the framework of wavefunction based methods like MRCI and/or
CASSCF can be evaluated:
via the QDPT approach within an individual Kramers doublet. This is valid analysis only for non-
integer spin cases. In particular for systems with well isolated Kramers doublets where the EPR
spectrum originates only from one Kramers doublet defined within the pseudo spin 1/2 formalism.
This analysis has been proven useful in determining the sign of the ZFS and the electronic structure of
the system under investigation. [389]
within the effective Hamiltonian approach. This is a valid analysis for all spin cases as it provides
the principal g-values of the system under investigation evaluated in the molecular axis frame. These
g-values can be directly compared with the experimentally determined ones. [390]
within the sum over states formalism (SOS). As above this analysis is valid for all spin cases and is
only available via the MRCI module.
9.23.2.4 Magnetization and Magnetic Susceptibility
The MRCI and CASSCF modules of ORCA allow for the calculation of magnetization and magnetic
susceptibility curves at different fields and temperatures by differentiation of the QDPT Hamiltonian with
respect to the magnetic field. For magnetic susceptibility, calculations are performed in two ways when a
static field different from zero is defined: (i) as the second derivative of energy with respect to the magnetic
field and (ii) as the magnetization divided by the magnetic field. Although the first method corresponds to
the definition of magnetic susceptibility, the second approach is widely used in the experimental determination
of T curves. If the static field is low, both formulas tend to provide similar values.
The full list of keywords is presented below.
%mrci
citype mrci
newblock 3 *
excitations none
refs cas(2,7) end
end
soc
dosoc true
domagnetization true # Calculate magnetization (def: false)
dosusceptibility true # Calculate susceptiblity (def: false)
LebedevPrec 5 # Precision of the grid for different field
# directions (meaningful values range from 1
# (smallest) to 10 (largest))
nPointsFStep 5 # number of steps for numerical differentiation
# (def: 5, meaningful values are 3, 5 7 and 9)
MAGFieldStep 100.0 # Size of field step for numerical differentiation
# (def: 100 Gauss)
MAGTemperatureMIN 4.0 # minimum temperature (K) for magnetization
MAGTemperatureMAX 4.0 # maximum temperature (K) for magnetization
MAGTemperatureNPoints 1 # number of temperature points for magnetization
MAGFieldMIN 0.0 # minimum field (Gauss) for magnetization

MAGFieldMAX 70000.0 # maximum field (Gauss) for magnetization
MAGNpoints 15 # number of field points for magnetization
SUSTempMIN 1.0 # minimum temperature (K) for susceptibility
SUSTempMAX 300.0 # maximum temperature (K) for susceptibility
SUSNPoints 300 # number of temperature points for susceptibility
SUSStatFieldMIN 0.0 # minimum static field (Gauss) for susceptibility
SUSStatFieldMAX 0.0 # maximum static field (Gauss) for susceptibility
SUSStatFieldNPoints 1 # number of static fields for susceptibility
end
end
The same keywords apply for CASSCF calculations in rel block (instead of soc in MRCI). Although different
aspects of integration and grid precision can be modified through keywords, default values should provide an
accurate description of both properties. Calculated magnetization and susceptibility are printed in .sus and
.mag files, respectively and also in the output file.
-------------------------------------------------------------------------------
FIELD DEPENDENT MAGNETIZATION AND MEAN SUSCEPTIBILITY (chi=M/B)
-------------------------------------------------------------------------------
TEMPERATURE (K) M. FIELD (Gauss) MAGNETIZATION (B.M.) chi*T (cm3*K/mol)
-------------------------------------------------------------------------------
4.00 0.00 0.000000 inf

4.00 5000.00 0.350759 1.567189
4.00 10000.00 0.688804 1.538788
4.00 15000.00 1.003466 1.494496
4.00 20000.00 1.287480 1.438115
4.00 25000.00 1.537346 1.373773
4.00 30000.00 1.752841 1.305282
4.00 35000.00 1.936067 1.235764
4.00 40000.00 2.090450 1.167516
4.00 45000.00 2.219920 1.102067
4.00 50000.00 2.328368 1.040315
4.00 55000.00 2.419335 0.982690
4.00 60000.00 2.495883 0.929301
4.00 65000.00 2.560582 0.880052
4.00 70000.00 2.615538 0.834730
-----------------------------------------------------------
-----------------------------------------------------------
TEMPERATURE DEPENDENT MAGNETIC SUSCEPTIBILITY
-----------------------------------------------------------
STATIC FIELD TEMPERATURE chi*T (cm3*K/mol)
(Gauss) (K) M/B d2E/dB2
-----------------------------------------------------------
0.00 1.00 ---- 1.576836

0.00 2.00 ---- 1.576910
0.00 3.00 ---- 1.576951
0.00 4.00 ---- 1.576988

0.00 5.00 ---- 1.577023
0.00 6.00 ---- 1.577057
0.00 7.00 ---- 1.577091
0.00 8.00 ---- 1.577125
0.00 9.00 ---- 1.577159
0.00 10.00 ---- 1.577193
0.00 11.00 ---- 1.577227
.....
0.00 300.00 ---- 1.586942
1000.00 1.00 1.570517 1.558042
1000.00 2.00 1.575324 1.572178
1000.00 3.00 1.576246 1.574845
1000.00 4.00 1.576590 1.575802
1000.00 5.00 1.576768 1.576264
1000.00 6.00 1.576880 1.576530
1000.00 7.00 1.576961 1.576704
1000.00 8.00 1.577026 1.576829
.....
9.23.2.5 MCD and Absorption Spectra
The MRCI module of the ORCA program allows calculating MCD spectra and the SOC effects on absorption
spectra. The formalism is described in detail by Ganyushin and Neese [391]. The approach is based on the
direct calculation of the transition energies and transition probabilities between the magnetic levels. Namely,
the differential absorption of LCP- and RCP photons for transitions from a manifold of initial states A to a
manifold of final states J. Using Fermis golden rule, the Franck-Condon approximation, assuming a pure
electronic dipole mechanism and accounting for the Boltzmann populations of the energy levels, the basic
equation of MCD spectroscopy may be written as (atomic units are used throughout):
X
2 2

= (Na Nj ) |ha |mLCP | j i| |ha |mRCP | j i| f (E) (9.292)
E a,j
Here a and j label members of the initial and state manifold probed in the experiments.
exp (Ea /kT )

Na (B, T ) = P (9.293)
exp (Ei /kT )
i
denotes the Boltzmann population and if the a-th ground state sublevel at energy Ea , f (E) stands for a line
shape function, and denotes a collection of constants. The electric dipole operators are given by:
mLCP mx imy (9.294)
mRCP mx + imy (9.295)

They represent linear combinations of the dipole moment operator:
X X
m
~ = ~N
ZN R ~ri (9.296)
N i
where N and i denotes summations of nuclei (at positions R ~ N with charges ZN ) and electrons (at positions
~ri ) respectively. The calculated transition dipole moment are subjected to the space averaging over the Euler
angles which is performed by a simple summation over three angular grids.
Z2 Z2 Z X
1
= sin ddd sin (9.297)
E ev 8 2 E
E
=0 =0 =0
Finally, every transition is approximated by a Gaussian curve with a definite Gaussian shape width parameter.
Hence, the final calculated MCD spectrum arises from the superposition of these curves.
As an illustration, consider calculation of a classical example of MCD spectrum of [Fe(CN)6 ]3 . The mrci
block of the input file is presented below.
%mrci ewin -4,10000

citype mrddci2
intmode ritrafo
Tsel 1e-6
Tpre 1e-5
etol 1e-8
rtol 1e-8
cimode direct2
maxmemint 300
solver diis
davidsonopt 0
nguessmat 150
MaxIter 50
LevelShift 0.5
PrintLevel 3
soc
printlevel 3
mcd true # perform the MCD calculation
NInitStates 24 # number of SOC and SSC state to account
# Starts from the lowest state
NPointsTheta 10 # number of integration point for
NPointsPhi 10 # Euler angles
NPointsPsi 10 #
B 43500 # experimental magnetic field strength
# in Gauss
Temperature 299.0 # experimental temperature (in K)
end
newblock 2 *
nroots 12
excitations cisd
refs cas(23,12) end
end
end
The parameters B and Temperature can be assigned in pairs, i.e. B = 1000, 2000, 3000. . . , Temperature = 4,
10, 300. . . . The program calculates the MCD and absorption spectra for every pair. Now for every point
of the integration grid the program prints out the Euler angles, the orientation of the magnetic field in the
coordinate system of a molecule, and the energy levels.
Psi = 36.000 Phi = 72.000 Theta = 20.000
Bx = 8745.0 By = 12036.5 Bz = 40876.6
Energy levels (cm-1,eV):Boltzmann populations for T = 299.000 K

0 : 0.000 0.0000 4.53e-01
1 : 3.943 0.0005 4.45e-01
2 : 454.228 0.0563 5.09e-02
3 : 454.745 0.0564 5.08e-02
4 : 1592.142 0.1974 2.13e-04
5 : 1595.272 0.1978 2.10e-04
6 : 25956.363 3.2182 2.59e-55
7 : 25958.427 3.2184 2.56e-55
8 : 25985.656 3.2218 2.25e-55
9 : 25987.277 3.2220 2.23e-55
10 : 26070.268 3.2323 1.49e-55
11 : 26071.484 3.2325 1.49e-55
12 : 31976.645 3.9646 6.78e-68
13 : 31979.948 3.9650 6.67e-68
14 : 32018.008 3.9697 5.56e-68
15 : 32021.074 3.9701 5.48e-68
16 : 32153.427 3.9865 2.90e-68
17 : 32157.233 3.9870 2.84e-68
18 : 42299.325 5.2444 1.81e-89
19 : 42303.461 5.2450 1.78e-89
20 : 42346.521 5.2503 1.45e-89
21 : 42348.023 5.2505 1.44e-89
22 : 42456.119 5.2639 8.53e-90
23 : 42456.642 5.2640 8.51e-90
In the next lines, ORCA calculates the strength of LCP and RCP transitions and prints the transition energies,
the difference between LCP and RCP transitions (denoted as C), and sum of LCP and RCP transitions
(denoted as D), and C by D ratio.
dE Na C D C/D
0 -> 1 3.943 4.53e-01 1.14e-13 8.13e-13 0.00e+00

0 -> 2 454.228 4.53e-01 5.01e-09 9.90e-09 5.06e-01
0 -> 3 454.745 4.53e-01 -4.65e-09 7.00e-09 -6.65e-01
0 -> 4 1592.142 4.53e-01 -8.80e-08 1.02e-07 -8.67e-01
0 -> 5 1595.272 4.53e-01 -2.29e-08 2.97e-08 -7.71e-01
0 -> 6 25956.363 4.53e-01 1.22e+01 9.60e+01 1.27e-01
0 -> 7 25958.427 4.53e-01 3.44e+01 3.52e+01 9.77e-01
0 -> 8 25985.656 4.53e-01 3.83e+01 1.70e+02 2.25e-01
0 -> 9 25987.277 4.53e-01 -7.73e+00 6.03e+01 -1.28e-01
0 ->10 26070.268 4.53e-01 -6.11e+00 2.85e+01 -2.14e-01
0 ->11 26071.484 4.53e-01 6.17e+00 9.21e+00 6.70e-01
0 ->12 31976.645 4.53e-01 2.45e+01 6.21e+01 3.95e-01
0 ->13 31979.948 4.53e-01 -6.58e+01 6.93e+01 -9.50e-01
0 ->14 32018.008 4.53e-01 3.42e-01 1.07e+02 3.21e-03
0 ->15 32021.074 4.53e-01 -6.16e+00 3.24e+01 -1.90e-01
0 ->16 32153.427 4.53e-01 -4.73e+01 1.37e+02 -3.46e-01
0 ->17 32157.233 4.53e-01 -1.02e+00 5.97e+01 -1.71e-02
0 ->18 42299.325 4.53e-01 6.47e+00 2.11e+01 3.07e-01
0 ->19 42303.461 4.53e-01 -2.59e+00 7.61e+00 -3.40e-01
0 ->20 42346.521 4.53e-01 1.90e+01 8.99e+01 2.11e-01
0 ->21 42348.023 4.53e-01 3.36e+00 3.55e+00 9.48e-01
0 ->22 42456.119 4.53e-01 2.52e-01 4.86e-01 5.20e-01
0 ->23 42456.642 4.53e-01 -2.01e+00 2.91e+00 -6.91e-01
1 -> 2 450.285 4.45e-01 4.59e-09 6.87e-09 6.69e-01
1 -> 3 450.802 4.45e-01 -4.96e-09 9.73e-09 -5.09e-01
All C and D values are copied additionally into the text files input.1.mcd, input.2.mcd. . . , for every pair of
Temperature and B parameters. These files contain the energies and C and D values for every calculated
transition. These files are used by the program orca mapspc to calculate the spectra lines. The orca mapspc
program generates from the raw transitions data into spectra lines. The main parameters of the orca mapspc
program are described in section 7.18.1. A typical usage of the orca mapspc program for MCD spectra
calculation for the current example may look as the following:
orca_mapspc input.1.mcd MCD -x020000 -x150000 -w2000
Here the interval for the spectra generation is set from 20000 cm1 to 50000 cm1 , and the line shape
parameter is set to 2000 cm1 .
Very often, it is desirable to assign different line width parameters to different peaks of the spectra to obtain
a better fitting to experiment. orca mapspc can read the line shape parameters from a simple text file named
as input.1.mcd.inp. This file should contain the energy intervals (in cm1 ) and the line shape parameters for
this energy interval in the form of:
20000 35000 1000

35000 40000 2000
40000 50000 1000
This file should not be specified in the executing command; orca mapspc checks for its presence automati-
cally:
orca_mapspc input.1.mcd MCD -x020000 -x150000

Mode is MCD
Number of peaks ... 276001
Start wavenumber [cm-1] ... 20000.0
Stop wavenumber [cm-1] ... 50000.0
Line width parameters are taken from the file:input.1.mcd.inp
Number of points ... 1024
Finally, the orca mapspc program generates the output text file input.1.mcd.dat which contains seven columns
of numbers: transition energies, intensities of MCD transitions (the MCD spectrum), intensities of absorption
transitions (the absorption spectrum), the ratio between the MCD and absorption intensities, and the last
three columns represent the sticks of the corresponding transitions.
Energy C D C/D C D E/D

24310.8 0.6673 980.2678 0.0006 0.0000 0.0000 0.0000
24340.1 0.8471 1174.3637 0.0007 -0.0001 0.0129 -0.0112
24369.5 1.0664 1408.5788 0.0007 0.0001 0.0281 0.0033
24398.8 1.3325 1690.5275 0.0007 0.0000 0.0000 0.0000
24428.1 1.6542 2029.0152 0.0008 0.0000 0.0000 0.0000
24457.4 2.0416 2434.1699 0.0008 0.0000 0.0332 0.0003
Now the MCD and the absorption spectra can be plotted with a suitable graphical program, for instance
with the Origin program.
9.23.2.6 Addition of Magnetic Fields
The inclusion of the Zeeman contribution into the QDPT procedure allows to obtain the splittings of the
magnetic levels in an external magnetic field. The switch for this calculation and the magnetic field strength
are defined in the soc subblock of the mrci block:
Figure 9.28: Calculated MCD and absorption spectra of [Fe(CN)6 ]3 (dash lines) compared to
experimental spectra (solid lines).
%mrci
soc
DoSOC true #
DoSSC true #
MagneticField true # default false
B 1,10,100,1000 # Strengh of the magnetic field in Gauss.
# 4000 is the default value
end
end
Then, the output contains three sets of data of splittings of the magnetic levels with the magnetic field
applied parallel to x, y, and z directions:
End B (Gauss) Energy levels (cm-1) and populations for B || x
1.0 -0.030 0.333 0.012 0.333 0.018 0.333

10.0 -0.030 0.333 0.012 0.333 0.018 0.333
100.0 -0.031 0.333 0.012 0.333 0.020 0.333
1000.0 -0.102 0.333 0.012 0.333 0.091 0.333
B (Gauss) Energy levels (cm-1) and populations for B || y
1.0 -0.030 0.333 0.012 0.333 0.018 0.333

10.0 -0.030 0.333 0.012 0.333 0.018 0.333
100.0 -0.032 0.333 0.014 0.333 0.018 0.333

1000.0 -0.105 0.334 0.018 0.333 0.087 0.333
B (Gauss) Energy levels (cm-1) and populations for B || z
1.0 -0.030 0.333 0.012 0.333 0.018 0.333

10.0 -0.030 0.333 0.011 0.333 0.018 0.333
100.0 -0.030 0.333 0.005 0.333 0.025 0.333
1000.0 -0.079 0.333 -0.030 0.333 0.108 0.333
Here the number in a row represents the strength of the magnetic field (in Gauss), and the following pairs of
numbers denote the energy of the magnetic level (in cm1 ) with its occupation number. This table can be
readily plotted with any suitable graphical program.
9.23.2.7 Relativistic Picture Change in Douglas-Kroll-Hess SOC and Zeeman Operators
The DKH correction to the SOC operator is implemented in ORCA as a correction to the one-electron part
of the SOMF operator. The DKH transformation is performed up to the second order, and the two-electron
part in our implementation is left untransformed. However, the electronic density employed for evaluating the
SOMF matrix elements is obtained from a scalar relativistic calculation. The inclusion of the DKH correction
is controlled by the picturechange key in the rel block:
%rel method DKH # relativistic method

picturechange 2 # include the DKH correction to SOC
end
The picturechange key can be set to 0, 1, and 2 for no picture change, the first order, and the second order
DKH transformations of the SOC operator.
With picturechange set to 1 or 2 the DKH correction are applied in the first order to the Zeeman operator.
This correction has a visible effect on calculated g-tensors for molecules containing third-row and heavier
atoms.
9.24 Multireference Equation of Motion Coupled-Cluster

(MR-EOM-CC) Theory
In analogy with single reference EOM-CC (see sections 8.7 and 9.20) and STEOM-CC (see sections 8.8 and
9.21), Multireference Equation of Motion Coupled-Cluster (MR-EOM-CC) theory [141146] can be viewed
a transform and diagonalize approach to molecular electronic structure theory. An MR-EOM calculation
involves a single state-averaged CASSCF calculation, incorporating a small number of low-lying states and the
solution of a single set of cluster amplitudes, which define a sequence of similarity transformed Hamiltonians.
The ultimate goal of these many-body transformations is to effectively decouple the CAS configurations from
important excited configurations (e.g. 2p2h, 2p1h, 1p1h, etc.) which comprise the first-order interacting
space. Through the definition of suitable cluster operators, in each of the transformations, the majority of
these excitations can be included in an internally-contracted fashion. Hence, the resulting final transformed
Hamiltonian can be diagonalized over a small subspace of the original first-order interacting space in order to
gain access to a large number of electronic states. As discussed in section 8.11, the MR-EOM implementation
in ORCA therefore, makes use of the CASSCF module (to obtain the state-averaged CASSCF reference), the
MDCI module for the solution of the amplitude equations and the calculation of the elements of the various
similarity transformed Hamiltonians and the MRCI module for the diagonalization of the final transformed
Hamiltonian. Some desirable features of this methodology are:
Many states can be obtained through the diagonalization of a similarity transformed Hamiltonian over
a compact diagonalization manifold (e.g. the final diagonalization space in MR-EOM-T|T |SXD|U
only includes the CAS configurations and 1h and 1p configurations).
Only a single state-averaged CASSCF calculation and the solution of a single set of amplitudes is
required to define the final similarity transformed Hamiltonian and the results are typically quite
insensitive to the precise definition of the CAS (only a few low lying multiplets need to be included in
the state-averaging)
The MR-EOM approach is rigourously invariant to rotations of the orbitals in the inactive, active and
virtual subspaces and it preserves both spin and spatial symmetry.
As the details concerning the MR-EOM methodology are rather involved, we refer the interested reader
to Refs. [141146] for a more detailed discussion. Note that the details concerning the implementation of
MR-EOM in ORCA can be found in Refs. [145] and [146]. In the following discussion, we note that general
spatial orbitals p, q, r, s, which comprise the molecular orbital basis, are partitioned into (doubly occupied)
inactive core orbitals i0 , j 0 , k 0 , l0 , occupied orbitals i, j, k, l (i.e. the union of the inactive core and active
orbital subspaces), active orbitals w, x, y, z and virtual orbitals a, b, c, d. In general, the many-body similarity
transformations assume the general form
b = eYb 1H
b Yb X X
gqp E
bp + pq b pq

G 2 e = g0 + q grs Ers + . . . , (9.298)
p,q p,q,r,s
in which Yb is a cluster operator and H2 is the bare Hamiltonian or a similarity transformed Hamiltonian
b
truncated up to two-body operators. The braces indicate Kutzelnigg-Mukherjee normal ordering [392, 393],
which is used extensively in the definition of the MR-EOM formalism. The various transformations which
need to be considered in the ORCA implementation of MR-EOM are summarized in Table 9.15. The
table also includes the expressions for the operator components of the various internally-contracted cluster
9.24 Multireference Equation of Motion Coupled-Cluster (MR-EOM-CC) Theory 593
Table 9.15: The details of the various MR-EOM transformations that are considered in the ORCA
implementation of MR-EOM. The equations for the operator components and the resid-
ual equations which determine the corresponding amplitudes also appear in the Table.
Note that we use the usual (Einstein) convention that repeated indices are summed
over.
Name Transformation Operators Operator Components Residual Equation
H = eT He
b b T
Tb1 = ta ba Ria = bi H
P
T Tb = Tb1 + Tb2 i Ei wm hm | Ea |m i
b b b
m
= h0 + h pq E
p
bq + h pq
pq 1 ab b ab ab = h ab
rs Ers + . . . Tb2 = t E Rij
b
2 ij ij ij
T b = eTb H
H
b Tb
2e Tb = Tb1 + Tb2 Tb1 = tia E
bi None (i.e. set tia ta
a i)
h pq E Tb2 =
p
h pq
pq 1 ij b ij
=e
h0 + e bq + e rs Ers + . . .
b t E
2 ab ab
None (i.e. set tij ab
ab tij )
b b b b 1 b S+
F = eS+X+D H2 e X+D b2 = saw b aw Riaw aw
b b b
SXD S
b=S
b2 S E
i0 j 0 i0 j 0 0 j 0 = f i0 j 0
= f 0 + f pq E
p
bq + f pq
pq
rs Ers + . . .
b X
b =X
b2 b2 = xaw
X b aw
E Riaw aw
0 x = f i0 x
i0 x i0 x
D
b =D
b2 b 2 = daw0 E
D b aw0 aw = f aw
Rxi0
xi xi xi0
U b = eUb F
G
b Ub
2e U
b=U
b2 b2 = uwx
U b wx
E Riwx wx
0 j 0 = gi0 j 0
i0 j 0 i0 j 0
= g0 + gqp E
p pq b pq
bq + grs Ers + . . .
operators and the residual equations that must be solved for the various amplitudes. Note that the residual
equations are typically of the many-body type (i.e. obtained by setting the corresponding elements of the
similarity transformed Hamiltonian to zero). The only exception is the residual equation which defines the tai
amplitudes, which is a projected equation of the form
X
Ria = bai H|

wm m |E b m . (9.299)
m

Here, |m is the mth state included in the state averaged CAS, with weight wm . The reason why the equation
for the singles is of the projected form is that it satisfies the Brillouin theorem (i.e. the first order singles
vanish for all i and a), whereas the corresponding many-body equation (h a = 0) does not.
i
Table 9.16: Details of the three MR-EOM approaches implemented in ORCA

Method Input Keyword Operators Diagonalization Manifold
MR-EOM-T|T -h-v MR-EOM-T|Td Tb; Tb CAS, 2h1p, 1h1p, 2h, 1h, 1p
MR-EOM-T|T |SXD-h-v MR-EOM-T|Td|SXD
Tb; Tb ; Sb + X b +D b CAS, 2h, 1h, 1p
MR-EOM-T|T |SXD|U-h-v MR-EOM b b b
T ; T ; S + X + D; U
b b b CAS, 1h, 1p
Note that there are three different MR-EOM approaches which have been implemented in ORCA. Namely, the
current implementation allows for MR-EOM-T|T -h-v, MR-EOM-T|T |SXD-h-v and MR-EOM-T|T |SXD|U-
h-v calculations. At this point it is useful to discuss the naming convention used for these approaches. We
use a vertical line to separate each transformation involved in the sequence of transformations defining the
given MR-EOM approach. For example T|T |SXD indicates that a T transformation, is followed by a T
tranformation, which is then followed by an SXD transformation. The h-v indicates that the elements of the
transformed Hamiltonian have been hermitized (h) and vertex symmetrized (v) before entering the MRCI
diagonalization (see Ref. [146] for more information). Essentially, this means that the full eightfold symmetry
of the two-electron integrals (and hermiticity of the one-body elements) have been enforced upon the elements
of the transformed Hamiltonian. The details of the three MR-EOM approaches are summarized in Table
9.16. This table includes the keyword (in the first line of input) used to initiate the calculation in ORCA,
the various operators involved and the configurations included in the final diagonalization manifold. One can
clearly see that the MR-EOM-T|T |SXD|U-h-v approach is the most cost effective, as it only includes the 1h
and 1p configurations, beyond the CAS, in the final diagonalization manifold.
The various %mdci keywords, which are important for controlling MR-EOM calculations are (i.e. default
values are given here):
%mdci
STol 1e-7 #Convergence Tolerance on Residual Equations
MaxIter 100 #Maximum Number of Iterations
DoSingularPT false #Activate the Singular PT/Projection Procedure
SingularPTThresh 0.01 #Threshold for the Singular PT/Projection
#Procedure
PrintOrbSelect false #Print the Eigenvalues of the Orbital Selection
#Densities (and R_core and R_virt values)
#and Terminate the Calculation
CoreThresh 0.0 #Core Orbital Selection Threshold
VirtualThresh 1.0 #Virtual Orbital Selection Threshold
end
As discussed below, the orbital selection scheme is activated by adding the keyword OrbitalSelection to
the first line of input. Keywords that are specific to the CASSCF and MRCI modules are discussed in sections
9.11 and 9.23, respectively. We note that in MR-EOM-T|T -h-v and MR-EOM-T|T |SXD-h-v calculations,
it is possible to overide the default excitation classes in the final MRCI diagonalization. This is done by
specifying excitations none and then explicitly setting the excitation flags within a given multiplicity block.
For example, if we wanted to have 1h, 1p, 1h1p, 2h and 2h1p excitations in the final diagonalization manifold,
we would specify (i.e. here we have requested 6 singlets and have a CAS(6,4) reference):
%mrci
newblock 1 *
excitations none
Flags[is] 1
Flags[sa] 1
Flags[ia] 1
Flags[ijss] 1
Flags[ijsa] 1
nroots 6
refs
cas(6,4)
end
end
end
9.24.1 The Steps Required to Run an MR-EOM Calculation
In order to illustrate the various steps required in a typical MR-EOM calculation, we will consider the
calculation of the excitation energies of the neutral Fe atom at the MR-EOM-T|T |SXD|U-h-v level of
theory.
9.24.1.1 State-Averaged CASSCF Calculation
Evidently, the first step is to determine a suitable state-averaged CASSCF reference for the subsequent
MR-EOM calculation. In choosing the state-averaged CAS for an MR-EOM calculation, we typically include
a few of the low-lying multiplets that have the same character as the (much larger number of) states that we
wish to compute in the final MR-EOM calculation. For the neutral Fe atom, we typically have electronic states
which have either 4s2 3d6 character or 4s1 3d7 character. From the NIST atomic spectra database [394,395], we
find that the lowest lying a5 D multiplet is of 4s2 3d6 character and the higher lying a5 F multiplet is of 4s1 3d7
character. Hence, we can set up a state-averaged CASSCF(8,6) calculation (i.e. 8 electrons in 6 orbitals (4s
and 3d)) which includes the 5 D and 5 F states and choose the weights such that the average occupation of
the 4s orbital is 1.5. As discussed in Ref. [147], this is done to avoid a preference toward either of the 4s
configurations in the state-averaging. We will run the state-averaged CASSCF calculation, making use of
the second order DKH (see 9.14.3) method for the inclusion of relativistic effects in a Def2-TZVPP basis
(i.e. the DKH-Def2-TZVPP relativistically recontracted basis, listed in section 9.3.1). The input file for the
state-averaged CASSCF(8,6) calculation takes the form:
!CASSCF DKH-Def2-TZVPP VeryTightSCF DKH
%casscf
nel 8
norb 6
mult 5
nroots 12
weights[0] = 0.7, 0.7, 0.7, 0.7, 0.7, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5
end
* xyz 0 5
Fe 0.000000 0.000000 0.000000
end
Here, we have requested 12 quintet states (the lowest lying 5 D and 5 F multiplets) and we have chosen the
weights to be 0.7 for the five 5 D states and 0.5 for the seven 5 F states, such that the overall occupation of the
4s orbital will be 1.5. Once the calculation has converged, it is important to inspect the results printed in the
final macro-iteration of the CASSCF calculation (macro-iteration 19 in this case). In this case, we have:
MACRO-ITERATION 19:
CI-ITERATION 0:
-1271.258898198 0.000000000000 ( 0.00)
-1271.258898198 0.000000000000
-1271.258898198 0.000000000000
-1271.258898198 0.000000000000
-1271.258898198 0.000000000000
-1271.186289789 0.000000000000
-1271.186289789 0.000000000000
-1271.186289789 0.000000000000
-1271.186289789 0.000000000000
-1271.186289789 0.000000000000
-1271.186289789 0.000000000000
-1271.186289789 0.000000000000
CI-PROBLEM SOLVED
DENSITIES MADE
E(CAS)= -1271.222593993 Eh DE= 0.000000000
N(occ)= 1.50000 1.30000 1.30000 1.30000 1.30000 1.30000
||g|| = 0.000000404 Max(G)= -0.000000204 Rot=18,5
Directly below CI-ITERATION 0, the final CAS-CI energies are printed and one observes that they follow
the correct degeneracy pattern (i.e. 5 states with energy -1271.258898198 and 7 states with energy
-1271.186289789). Furthermore, the final state-averaged CASSCF energy (E(CAS)= -1271.222593993) and
occupation numbers (N(occ)= 1.50000 1.30000 1.30000 1.30000 1.30000 1.30000) are also printed.
As expected, the occupation number of the 4s orbital is indeed 1.5, while the 3d orbitals each have an
occupation of 1.3.
9.24.1.2 Selection of the States to Include in the MR-EOM Calculation
Once a satisfactory CASSCF reference has been obtained, the next step is to determine the number of states
to include in the MR-EOM calculation. From the NIST atomic spectra database, one finds that the higher
lying states of 4s2 3d6 and 4s1 3d7 character are either singlets, triplets or quintets. In order to figure out how
many states should be included in each multiplicity block, one can perform an inexpensive CAS-CI calculation.
This is done by reading in the orbitals from the previous CASSCF calculation (here they are stored in the
file CAS.gbw) and requesting a single iteration (i.e. using the NoIter keyword) of a state-averaged CASSCF
calculation:
!CASSCF DKH-Def2-TZVPP ExtremeSCF DKH NoIter
!MOREAD
%moinp "CAS.gbw"
%casscf
nel 8
norb 6
mult 5,3,1
nroots 15,90,55
end
* xyz 0 5
Fe 0.000000 0.000000 0.000000
end
Here, after some experimentation, we have chosen 15 quintets, 90 triplets and 55 singlets. It is important that
we calculate states up to sufficiently high energy (i.e. all the states that are of interest) and it is absolutely
imperative that we have complete multiplets. Hence, several iterations of this procedure might be required to
choose the proper number of states for each multiplet. The relevant section of the output file which should
be analyzed is the SA-CASSCF TRANSITION ENERGIES. For the above calculation, we obtain (i.e. only the
CAS-CI energies for the first 33 roots are shown here):
-----------------------------
------------------------------
LOWEST ROOT (ROOT 0 ,MULT 5) = -1271.258898198 Eh -34592.713 eV
STATE ROOT MULT DE/a.u. DE/eV DE/cm**-1

1: 1 5 0.000000 0.000 0.0
2: 2 5 0.000000 0.000 0.0
3: 3 5 0.000000 0.000 0.0
4: 4 5 0.000000 0.000 0.0
5: 5 5 0.072608 1.976 15935.7
6: 6 5 0.072608 1.976 15935.7
7: 7 5 0.072608 1.976 15935.7
8: 8 5 0.072608 1.976 15935.7
9: 9 5 0.072608 1.976 15935.7
10: 10 5 0.072608 1.976 15935.7
11: 11 5 0.072608 1.976 15935.7
12: 0 3 0.092859 2.527 20380.1
13: 1 3 0.092859 2.527 20380.1
14: 2 3 0.092859 2.527 20380.1
15: 3 3 0.092859 2.527 20380.1
16: 4 3 0.092859 2.527 20380.1
17: 5 3 0.092859 2.527 20380.1
18: 6 3 0.092859 2.527 20380.1
19: 8 3 0.092859 2.527 20380.1
20: 7 3 0.092859 2.527 20380.1
21: 9 3 0.092859 2.527 20380.1
22: 10 3 0.092859 2.527 20380.1
23: 11 3 0.101847 2.771 22352.7
24: 12 3 0.101847 2.771 22352.7
25: 13 3 0.101847 2.771 22352.7
26: 14 3 0.101847 2.771 22352.7
27: 15 3 0.101847 2.771 22352.7
28: 16 3 0.101847 2.771 22352.7
29: 17 3 0.101847 2.771 22352.7

30: 18 3 0.102559 2.791 22509.1
31: 19 3 0.102559 2.791 22509.1
32: 20 3 0.102559 2.791 22509.1
9.24.1.3 Running the MR-EOM Calculation
Now that we have chosen a suitable CASSCF reference and the states that we wish to calculate, we can
finally proceed with the MR-EOM calculation. The following input file runs an MR-EOM-T|T |SXD|U-h-v
calculation for 15 quintet, 90 triplet and 55 singlet states of the neutral Fe atom (i.e. the CASSCF orbitals
are read from CAS.gbw):
!MR-EOM DKH-Def2-TZVPP VeryTightSCF DKH
!MOREAD
%moinp "CAS.gbw"
%method frozencore fc_ewin end
%casscf
nel 8
norb 6
mult 5
nroots 12
weights[0] = 0.7, 0.7, 0.7, 0.7, 0.7, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5
end
%mdci
ewin -6, 10000
STol 1e-7
end
%mrci
ewin -6, 10000
MaxIter 200
newblock 5 *
nroots 15
refs cas(8,6) end
end
newblock 3 *
nroots 90
refs cas(8,6) end
end
newblock 1 *
nroots 55
refs cas(8,6) end
end
end
* xyz 0 5
Fe 0.000000 0.000000 0.000000
end
Note that since the default frozen core settings exclude the 3p orbitals from the correlation treatment, we
have used an energy window (i.e. the line ewin -6, 10000 in both the %mdci and %mrci blocks) such that
they are included in the current calculation. We note that a detailed discussion of the input and output of an
MR-EOM calculation has already been given in section 8.11 and thus, we do not repeat it here. It is important
to reiterate that one should always inspect the values of the largest (T, S and U) amplitudes. Ideally, the
largest amplitudes should be smaller than 0.1 and should not exceed 0.15. If some of the amplitudes are
larger than 0.15, it might be necessary to revisit the definition of the CAS and the weights used. For the T
amplitudes, an alternative solution is to use the projection/singular PT scheme discussed in section 9.24.3
below.
As discussed in section 8.11, the excitation energies are printed under the heading TRANSITION ENERGIES.
For the current calculation, we obtain the following results (only the results for 33 states are shown here):
-------------------
TRANSITION ENERGIES
-------------------

0 5 -1 0 0 0.000 0.000 0.0
1 5 -1 1 0 0.000 0.000 0.0
2 5 -1 2 0 0.000 0.000 0.0
3 5 -1 3 0 0.000 0.000 0.0
4 5 -1 4 0 0.000 0.000 0.0
5 5 -1 5 0 33.901 0.922 7440.3
6 5 -1 6 0 33.901 0.922 7440.3
7 5 -1 7 0 33.901 0.922 7440.3
8 5 -1 8 0 33.901 0.922 7440.3
9 5 -1 9 0 33.901 0.922 7440.3
10 5 -1 10 0 33.901 0.922 7440.3
11 5 -1 11 0 33.901 0.922 7440.3
12 3 -1 0 1 54.743 1.490 12014.7
13 3 -1 1 1 54.743 1.490 12014.7
14 3 -1 2 1 54.743 1.490 12014.7
15 3 -1 3 1 54.743 1.490 12014.7
16 3 -1 4 1 54.743 1.490 12014.7
17 3 -1 5 1 54.743 1.490 12014.7

18 3 -1 6 1 54.743 1.490 12014.7
19 5 -1 12 0 78.790 2.144 17292.4
20 5 -1 13 0 78.790 2.144 17292.4
21 5 -1 14 0 78.790 2.144 17292.5
22 3 -1 7 1 95.413 2.596 20940.8
23 3 -1 8 1 95.413 2.596 20940.8
24 3 -1 9 1 95.413 2.596 20940.8
25 3 -1 10 1 95.413 2.596 20940.8
26 3 -1 11 1 95.413 2.596 20940.8
27 3 -1 12 1 95.413 2.596 20940.8
28 3 -1 13 1 95.413 2.596 20940.8
29 3 -1 14 1 95.413 2.596 20940.8
30 3 -1 15 1 95.413 2.596 20940.8
31 3 -1 16 1 95.413 2.596 20940.8
32 3 -1 17 1 95.413 2.596 20940.8
It is also important to recall that one should always inspect the reference weights for each state, as only
states which are dominated by reference space configurations can be treated accurately at the MR-EOM level
of theory. As a general rule, the reference weights should be larger (or close to) 0.9. In each multiplicity
block, the individual state energies and reference weights can be found following convergence of the MRCI
procedure, under the heading CI-RESULTS (see section 8.11 for a more detailed discussion).
9.24.2 Approximate Inclusion of Spin-Orbit Coupling Effects in MR-EOM

Calculations
The effects of spin-orbit coupling can approximately be included in MR-EOM calculations using the SOC
submodule of the MRCI module, as outlined in section 9.23.2. This can be viewed as a first order approximation
to the inclusion of spin-orbit coupling effects in MR-EOM. In a more rigourous formulation, one would have
to consider the various similarity transformations of the spin-orbit coupling operator. The details of the
SOC submodule of the MRCI module have already been discussed in detail in 9.23.2 and its usage within
the MR-EOM formalism is identical to that discussed therein. Let us consider the calculation of spin-orbit
coupling effects in the excitation spectrum of the neutral Fe atom considered in the previous section. The
input file for this calculation is:
!MR-EOM DKH-Def2-TZVPP ExtremeSCF DKH
%method frozencore fc_ewin end
%casscf
nel 8
norb 6
mult 5
nroots 12
weights[0] = 0.7, 0.7, 0.7, 0.7, 0.7, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5
etol 1e-12
gtol 1e-12
end
%mdci
ewin -6, 10000
MaxIter 300
STol 1e-12
end
%mrci
ewin -6, 10000
MaxIter 200
newblock 5 *
nroots 15
refs cas(8,6) end
end
newblock 3 *
nroots 90
refs cas(8,6) end
end
newblock 1 *
nroots 55
refs cas(8,6) end
end
soc
dosoc true #include spin-orbit coupling effects
end
end
* xyz 0 5
Fe 0.000000 0.000000 0.000000
end
In contrast with the calculation performed in section 9.24.1, the convergence thresholds have been tightened in
all aspects of the calculation (i.e. the use of the ExtremeSCF keyword, etol and gtol (CASSCF energy and
orbital gradient convergence tolerance) are set to 1 1012 and the convergence tolerance for the residuals in
the MR-EOM amplitude iterations have been set to 1 1012 ). We note that with the use of the ExtremeSCF
keyword, the convergence tolerance on the energy (Etol) and residual (Rtol) in the MRCI portion of the
calculation are also set to 1 1012 . Although it is not absolutely necessary, we have used very strict
convergence thresholds in order to preserve the degeneracies of the various multiplets as much as possible.
The output of spin-orbit corrected MR-EOM spectrum appears under the heading SPIN-ORBIT CORRECTED
MRCI ABSORPTION SPECTRUM (i.e. only the excitation energies for the first 36 states are shown here):
------------------------------------------------------------------------------------------
SPIN-ORBIT CORRECTED MRCI ABSORPTION SPECTRUM
------------------------------------------------------------------------------------------
States Energy Wavelength fosc T2 |TX| |TY| |TZ|
(cm-1) (nm) (D**2) (D) (D) (D)
------------------------------------------------------------------------------------------
0 1 0.0 6262179496748.3 0.000000000 0.00000 0.00000 0.00000 0.00000
0 2 0.0 3643449889017.2 0.000000000 0.00000 0.00000 0.00000 0.00000

0 3 0.0 2456298495.9 0.000000000 0.00000 0.00000 0.00000 0.00000
0 4 0.0 2453471569.3 0.000000000 0.00000 0.00000 0.00000 0.00000
0 5 0.0 1842867662.6 0.000000000 0.00000 0.00000 0.00000 0.00000
0 6 0.0 1840768531.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 7 0.0 1840312096.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 8 0.0 1364058512.5 0.000000000 0.00000 0.00000 0.00000 0.00000
0 9 391.8 25525.6 0.000000000 0.00000 0.00000 0.00000 0.00000
0 10 391.8 25525.6 0.000000000 0.00000 0.00000 0.00000 0.00000
0 11 391.8 25525.6 0.000000000 0.00000 0.00000 0.00000 0.00000
0 12 391.8 25525.6 0.000000000 0.00000 0.00000 0.00000 0.00000
0 13 391.8 25525.6 0.000000000 0.00000 0.00000 0.00000 0.00000
0 14 391.8 25525.6 0.000000000 0.00000 0.00000 0.00000 0.00000
0 15 391.8 25525.6 0.000000000 0.00000 0.00000 0.00000 0.00000
0 16 671.9 14883.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 17 671.9 14883.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 18 671.9 14883.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 19 671.9 14883.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 20 671.9 14883.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 21 853.8 11711.8 0.000000000 0.00000 0.00000 0.00000 0.00000
0 22 853.8 11711.8 0.000000000 0.00000 0.00000 0.00000 0.00000
0 23 853.8 11711.8 0.000000000 0.00000 0.00000 0.00000 0.00000
0 24 943.7 10597.0 0.000000000 0.00000 0.00000 0.00000 0.00000
0 25 7317.2 1366.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 26 7317.2 1366.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 27 7317.2 1366.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 28 7317.2 1366.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 29 7317.2 1366.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 30 7317.2 1366.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 31 7317.2 1366.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 32 7317.2 1366.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 33 7317.2 1366.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 34 7317.2 1366.7 0.000000000 0.00000 0.00000 0.00000 0.00000
0 35 7317.2 1366.7 0.000000000 0.00000 0.00000 0.00000 0.00000
The first 25 values in the table above make up the a5 D states with J values from J = L + S = 2 + 2 =
4 to J = L - S = 0. The states 25 to 35 constitute the J = 5 level of the a5 F multiplet. Also, note that
the wavelengths that are printed for the first eight states have no physical meaning (i.e. correspond to zero
energy). The large values that appear in the output above should therefore, be disregarded.
It is possible to obtain more accurate results by performing an MR-EOM-T|T |SXD-h-v calculation and
including the 1h1p excitations. It is important to note that these calculations are significantly more expensive.
As discussed above, to run an MR-EOM-T|T |SXD-h-v calculation, the keyword MR-EOM-T|Td|SXD must
appear in the first line of input and in order to activate the 1h1p excitations in each multiplicity block of the
MRCI calculation, the %mrci block takes the form:
%mrci
ewin -6, 10000
MaxIter 200
newblock 5 *
nroots 15
excitations none
Flags[is] 1
Flags[sa] 1
Flags[ia] 1
Flags[ijss] 1
refs cas(8,6) end
end
newblock 3 *
nroots 90
excitations none
Flags[is] 1
Flags[sa] 1
Flags[ia] 1
Flags[ijss] 1
refs cas(8,6) end
end
newblock 1 *
nroots 55
excitations none
Flags[is] 1
Flags[sa] 1
Flags[ia] 1
Flags[ijss] 1
refs cas(8,6) end
end
soc
dosoc true
end
end
We use excitations none to set the default excitation flags to false and then manually set the 1h (Flags[is]),
1p (Flags[sa]), 1h1p (Flags[ia]) and 2h (Flags[ijss]) excitation flags to true.
WARNINGS
Currently, MR-EOM-T|T |SXD|U-h-v calculations can only be run with the default excitation classes
in the final MRCI (i.e. 1h and 1p). Any other input options for the excitation flags will automatically
be overwritten and set to the default values.
Only the inclusion of spin-orbit coupling effects has been tested for MR-EOM calculations. Other
features that are available in the MRCI module (e.g. spin-spin coupling, magnetic property calculations,
etc.) have not been tested for use within MR-EOM calculations.
9.24.3 A Projection/Singular PT Scheme to Overcome Convergence Issues in the T

Amplitude Iterations
In certain cases, there may be nearly singular T2 amplitudes (often, but not always large in magnitude),
which can cause convergence issues in the solution of the T amplitude equations. Hence, it is sometimes
necessary to discard some of the amplitudes in order to remedy these convergence problems. The nearly
singular T2 amplitudes are of the form tab wx , where (w, x) is a pair of active orbitals which corresponds to
a small eigenvalue (pair occupation number nwx ) of the two-body reduced density matrix (RDM). When
nearly singular amplitudes are present, it is possible to employ a singular PT/projection scheme (i.e. Scheme
I described in Ref. [141]), using the two-body RDM as the metric matrix, in order to discard these nearly
singular amplitudes and replace them with suitable perturbative estimates. As a first example, lets consider
the following calculation on the cyclopentadiene molecule:
!MR-EOM def2-SVP VeryTightSCF
%casscf
nel 4
norb 4
nroots 2
mult 3
end
%mdci
STol 1e-7
MaxIter 60
end
%mrci
newblock 1 *
nroots 3
refs cas(4,4) end
end
newblock 3 *
nroots 3
refs cas(4,4) end
end
end
* xyz 0 1
H -0.879859 0.000000 1.874608
H 0.879859 0.000000 1.874608
H 0.000000 2.211693 0.612518
H 0.000000 -2.211693 0.612518
H 0.000000 1.349811 -1.886050
H 0.000000 -1.349811 -1.886050
C 0.000000 0.000000 1.215652
C 0.000000 -1.177731 0.285415
C 0.000000 1.177731 0.285415

C 0.000000 -0.732372 -0.993420
C 0.000000 0.732372 -0.993420
*
The T amplitude iterations do not converge after 60 iterations and show no signs of convergence (i.e. final
largest residual of 0.000458135 and oscillatory behaviour over a significant portion of the iterations). If we
inspect the largest T amplitudes,
--------------------
--------------------
19-> 24 19-> 24 0.043128
19-> 23 19-> 23 0.031123
11-> 25 11-> 25 0.028458
19-> 41 19-> 41 0.027950
11-> 47 11-> 47 0.027026
19-> 22 19-> 22 0.025188
15-> 26 15-> 26 0.022084
19-> 21 19-> 21 0.022038
11-> 47 11-> 25 0.022033
11-> 25 11-> 47 0.022033
19-> 29 19-> 24 0.021769
19-> 24 19-> 29 0.021769
19-> 36 19-> 36 0.020987
17-> 38 17-> 38 0.019743
19-> 41 16-> 36 0.019107
18-> 40 18-> 40 0.017949
one can see that there are no unusually large amplitudes. If we turn on the singular PT/projection scheme
by adding the line DoSingularPT true to the %mdci block:
%mdci
STol 1e-7
MaxIter 60
DoSingularPT true
end
and rerun the calculation, we find that the T amplitude iterations now successfully converge in 23 iterations.
If we look at the largest T amplitudes:
--------------------
--------------------
11-> 25 11-> 25 0.028440
11-> 47 11-> 47 0.027027
15-> 26 15-> 26 0.022069
11-> 47 11-> 25 0.022031
11-> 25 11-> 47 0.022031

19-> 41 19-> 41 0.020463
17-> 38 17-> 38 0.018288
11-> 43 11-> 43 0.017250
11-> 39 11-> 39 0.016838
15-> 27 15-> 27 0.016001
13-> 26 13-> 26 0.015985
16-> 36 16-> 36 0.015759
19-> 41 16-> 36 0.015697
18-> 40 18-> 40 0.015376
17-> 31 17-> 31 0.015074
18-> 40 17-> 38 0.014470
the majority of the amplitudes corresponding to the active pair (w, x) = (19, 19) no longer appear in the list
(i.e. they are nearly singular amplitudes which have been projected out). The only one that does appear in
the list, corresponds to a projected perturbative estimate (e.g. 19-> 41 19-> 41 0.020463).
By default, when the singular PT/projection scheme is active, the amplitudes tab
wx for which the pair occupation
numbers satisfy nwx < 0.01 (i.e. SingularPTThresh =0.01), are replaced by perturbative amplitudes in
the procedure. However, in some cases, it might be necessary to increase the SingularPTThresh threshold
beyond the default value in order to achieve convergence. One such example is the ferrocene molecule.
Consider the following calculation:
!MR-EOM def2-SVP
%casscf
nel 6
norb 5
mult 1,3
nroots 5,6
end
%mdci
DoSingularPT true
MaxIter 50
end
%mrci
newblock 1 *
nroots 18
refs cas(6,5) end
end
newblock 3 *
nroots 10
refs cas(6,5) end
end
end
* xyz 0 1
Fe 0.000000 0.000000 0.000000
C 0.000000 1.220080 1.650626
C -1.160365 0.377025 1.650626
C -0.717145 -0.987065 1.650626
C 0.717145 -0.987065 1.650626
C 1.160365 0.377025 1.650626
C 0.000000 1.220080 -1.650626
C 1.160365 0.377025 -1.650626
C 0.717145 -0.987065 -1.650626
C -0.717145 -0.987065 -1.650626
C -1.160365 0.377025 -1.650626
H 0.000000 2.306051 1.635648
H -2.193184 0.712609 1.635648
H -1.355463 -1.865634 1.635648
H 1.355463 -1.865634 1.635648
H 2.193184 0.712609 1.635648
H 0.000000 2.306051 -1.635648
H 2.193184 0.712609 -1.635648
H 1.355463 -1.865634 -1.635648
H -1.355463 -1.865634 -1.635648
H -2.193184 0.712609 -1.635648
end
The T amplitude iterations do not converge after 50 iterations, even though the singular PT/projection
scheme is activated. If we increase SingularPTThresh to 0.05 by adding the line SingularPTThresh 0.05
to the %mdci block:
%mdci
DoSingularPT true
SingularPTThresh 0.05
MaxIter 50
end
the T amplitude iterations successfully converge in 25 iterations.
In conclusion, it occasionally happens that the T amplitude iterations do not converge. In these cases,
the singular PT/projection scheme can be activated (DoSingularPT true) to overcome these convergence
difficulties. Sometimes, like in the case of ferrocene, it is necessary to adjust the threshold for the singular
PT/projection procedure (SingularPTThresh) in order to achieve convergence. If the procedure still fails
with larger values of the threshold, then it might be necessary to revisit the definition of the state-averaged
CAS.
9.24.4 An Orbital Selection Scheme for More Efficient Calculations of Excitation

Spectra with MR-EOM
As described in Ref. [145], the MR-EOM implementation in ORCA can make use of a sophisticated scheme
to discard inactive and virtual orbitals, which are not important for the description of the excited states
of interest. The selection of inactive core orbitals is based on the eigenvalues of the core orbital selection
density
Tr (Dt )
Di0 j 0 = Dit0 j 0 + Dis0 j 0 + Diu0 j 0 ,

(9.300)
Tr (Ds ) + Tr (Du )
in which
X (1)
(1) (1)

Dit0 j 0 = tab
i0 w 2tab ba
j 0 w tj 0 w , (9.301)
w,a,b
X h (1) (1) (1)
(1)
(1) (1)
i
Dis0 j 0 = saw
i0 k 2aw
j0k saw
kj 0 + saw
ki0 2saw
kj 0 saw
j0k , (9.302)
k,w,a
X (1)
(1) (1)

Diu0 j 0 = uwx
i0 k 0 2uwx wx
j 0 k0 uk0 j 0 , (9.303)
k0 ,w,x
(1) (1) (1)

are respectively, the contributions from the first order tab aw
i0 w , si0 k and uwx
i0 k0 amplitudes (i.e. note that
all amplitudes have at least one active label). Similarly, the selection of virtual orbitals is based upon the
eigenvalues of the virtual orbital selection density
Tr (t ) s
ab = tab + , (9.304)
Tr (s ) ab
in which, the contribution t , from the first order T2 amplitudes, is given by
X (1)
(1) (1)
X (1)
(1) (1)

tab = tac
wk 2tbc cb
wk twk + tac
i0 w 2tbc cb
i0 w t i0 w , (9.305)
k,w,c i0 ,w,c
and the contribution s , from the first order S2 amplitudes, is given by
X (1)
(1) (1)
X (1)
(1) (1)

sab = saw
i0 k 2sbw
i0 k sbw
ki0 + saw
xi0 2sbw
xi0 sbw
i0 x . (9.306)
i0 ,k,w i0 ,w,x
Diagonalization of the core orbital selection density Di0 j 0 and the virtual orbital selection density ab then
yields two respective sets of eigenvalues {i0 } and {a }. We have found it useful to compute the ratios,
Pnexcluded
core
i0 =0 i0
Rcore = P ncore 100%, (9.307)
i0 =0 i
0
Pnexcluded
virt
a=0 a
Rvirt = P nvirt 100%, (9.308)
a=0 a
of the sum of the excluded eigenvalues to the sum over all eigenvalues. The orbital selection in the core and
virtual subspaces is then based upon the values of these ratios, as will be discussed below.
The orbital selection procedure is activated by adding the keyword OrbitalSelection to the first line of
input, e.g.
! MR-EOM def2-TZVPP VeryTightSCF OrbitalSelection
There are two threshold parameters CoreThresh and VirtualThresh, which are used to determine which
inactive core and virtual orbitals are to be discarded in the orbital selection procedure, respectively. Namely,
all inactive core orbitals for which Rcore < CoreThresh (i.e. Rcore as defined in Eq. 9.307) are discarded
and all virtual orbitals satisfying the condition Rvirt < VirtualThresh (i.e. Rvirt as defined in Eq. 9.308)
are discarded. The default values of these thresholds are CoreThresh = 0.0 (no core orbital selection) and
VirtualThresh = 1.0. However, the values of these parameters can easily be changed by redefining them in
the %mdci block:
%mdci
CoreThresh 1.0
VirtualThresh 1.0
end
Let us consider the calculation of the previous section (9.24.3) on ferrocene, with the orbital selection
procedure activated (using the default thresholds):
!MR-EOM def2-SVP OrbitalSelection
%casscf
nel 6
norb 5
mult 1,3
nroots 5,6
end
%mdci
DoSingularPT true
MaxIter 50
end
%mrci
newblock 1 *
nroots 18
refs cas(6,5) end
end
newblock 3 *
nroots 10
refs cas(6,5) end
end
end
* xyz 0 1
Fe 0.000000 0.000000 0.000000
C 0.000000 1.220080 1.650626
C -1.160365 0.377025 1.650626
C -0.717145 -0.987065 1.650626
C 0.717145 -0.987065 1.650626
C 1.160365 0.377025 1.650626
C 0.000000 1.220080 -1.650626
C 1.160365 0.377025 -1.650626
C 0.717145 -0.987065 -1.650626
C -0.717145 -0.987065 -1.650626
C -1.160365 0.377025 -1.650626
H 0.000000 2.306051 1.635648
H -2.193184 0.712609 1.635648
H -1.355463 -1.865634 1.635648
H 1.355463 -1.865634 1.635648
H 2.193184 0.712609 1.635648
H 0.000000 2.306051 -1.635648
H 2.193184 0.712609 -1.635648
H 1.355463 -1.865634 -1.635648
H -1.355463 -1.865634 -1.635648
H -2.193184 0.712609 -1.635648
end
The details of the orbital selection procedure are printed under the heading ORBITAL SELECTION:
------------------------------------------------
ORBITAL SELECTION
------------------------------------------------
T1 is NOT used in the construction of the orbital selection densities

Factor (in percent) for inactive (core) orbital selection ... 0.000000000
Factor (in percent) for virtual orbital selection ... 1.000000000
Inactive orbitals before selection: 15 ... 44 ( 30 MOs/ 60 electrons)
Virtual orbitals before selection: 50 ... 220 (171 MOs )
Inactive orbitals after selection: 15 ... 44 ( 30 MOs/ 60 electrons)
Virtual orbitals after selection: 50 ... 126 ( 77 MOs )
-------------------------------------------
TIMINGS FOR THE ORBITAL SELECTION PROCEDURE
-------------------------------------------
Total Time for Orbital Selection ... 98.492 sec

First Half Transformation ... 93.948 sec ( 95.4%)

Second Half Transformation ... 2.752 sec ( 2.8%)
Formation of Orbital Selection Densities ... 1.775 sec ( 1.8%)
Core Orbital Selection ... 0.001 sec ( 0.0%)
Virtual Orbital Selection ... 0.009 sec ( 0.0%)
Finalization of Orbitals ... 0.007 sec ( 0.0%)
Comparing the number of virtual orbitals before the orbital selection procedure (171) with the number that are
left after orbital selection (77), we see that more than half have been discarded (94). The canonical calculation
(without orbital selection) takes 149373 seconds to run and yields the following excitation energies:
-------------------
TRANSITION ENERGIES
-------------------

0 1 -1 0 0 0.000 0.000 0.0
1 3 -1 0 1 65.110 1.772 14289.9
2 3 -1 1 1 65.110 1.772 14289.9
3 3 -1 2 1 70.413 1.916 15454.0
4 3 -1 3 1 70.413 1.916 15454.0
5 3 -1 4 1 95.979 2.612 21065.0
6 3 -1 5 1 95.979 2.612 21065.0
7 1 -1 1 0 105.302 2.865 23111.1
8 1 -1 2 0 105.302 2.865 23111.1
9 1 -1 3 0 107.034 2.913 23491.4
10 1 -1 4 0 107.034 2.913 23491.4
11 1 -1 5 0 160.595 4.370 35246.6
12 1 -1 6 0 160.596 4.370 35246.6
13 3 -1 6 1 164.694 4.482 36146.1
14 3 -1 7 1 165.379 4.500 36296.6
15 3 -1 8 1 165.379 4.500 36296.6
16 3 -1 9 1 171.464 4.666 37632.1
17 1 -1 7 0 208.587 5.676 45779.6
18 1 -1 8 0 208.587 5.676 45779.6
19 1 -1 9 0 213.093 5.799 46768.6
20 1 -1 10 0 213.093 5.799 46768.6
21 1 -1 11 0 216.225 5.884 47456.0
22 1 -1 12 0 220.230 5.993 48334.9
23 1 -1 13 0 220.230 5.993 48334.9
24 1 -1 14 0 224.583 6.111 49290.3
25 1 -1 15 0 224.583 6.111 49290.3
26 1 -1 16 0 237.914 6.474 52216.0
27 1 -1 17 0 237.914 6.474 52216.0
In contrast, the calculation with the orbital selection procedure activated runs in 28977 seconds (a factor of 5
speedup) and produces the following excitation energies:
-------------------
TRANSITION ENERGIES
-------------------

0 1 -1 0 0 0.000 0.000 0.0
1 3 -1 0 1 65.112 1.772 14290.4
2 3 -1 1 1 65.134 1.772 14295.3
3 3 -1 2 1 70.520 1.919 15477.3
4 3 -1 3 1 70.520 1.919 15477.3
5 3 -1 4 1 96.105 2.615 21092.7
6 3 -1 5 1 96.134 2.616 21099.0
7 1 -1 1 0 105.415 2.868 23136.0
8 1 -1 2 0 105.450 2.869 23143.5
9 1 -1 3 0 107.294 2.920 23548.3
10 1 -1 4 0 107.294 2.920 23548.3
11 1 -1 5 0 161.082 4.383 35353.4
12 1 -1 6 0 161.094 4.384 35356.0
13 3 -1 6 1 164.786 4.484 36166.4
14 3 -1 7 1 165.465 4.503 36315.4
15 3 -1 8 1 165.465 4.503 36315.5
16 3 -1 9 1 171.542 4.668 37649.1
17 1 -1 7 0 208.853 5.683 45838.0
18 1 -1 8 0 208.853 5.683 45838.0
19 1 -1 9 0 213.419 5.807 46840.1
20 1 -1 10 0 213.419 5.807 46840.1
21 1 -1 11 0 216.526 5.892 47521.9
22 1 -1 12 0 220.611 6.003 48418.4
23 1 -1 13 0 220.611 6.003 48418.5
24 1 -1 14 0 225.135 6.126 49411.5
25 1 -1 15 0 225.136 6.126 49411.5
26 1 -1 16 0 238.388 6.487 52320.1
27 1 -1 17 0 238.388 6.487 52320.1
We note that the excitation energies in the orbital selection procedure agree very nicely with those of
the canonical calculation. However, the total energies are significantly different, as we currently have not
implemented a procedure to correct them. Hence, the following warning is very important.
WARNING
The orbital selection procedure should only be used for the calculation of excitation energies. Total
energies computed with the orbital selection procedure have not been corrected and can differ greatly
from the canonical results.
Before leaving the discussion of the orbital selection procedure, we note that there is also a keyword
PrintOrbSelect, which can be added to the %mdci block in order to print the eigenvalues of the inactive
core orbital selection and virtual orbital selection densities and the corresponding values of Rcore and Rvirt
defined in Eqs. 9.307 and 9.308, respectively. This is useful if one wants to manually select the orbitals to
discard in the orbital selection procedure by adjusting the values of CoreThresh and VirtualThresh. We
note that the program terminates after printing. In the case of the calculation on ferrocene, if we modify the
%mdci block to read
%mdci
DoSingularPT true
MaxIter 50
PrintOrbSelect True
end
we find the following information in the ORBITAL SELECTION section of the output (only the first 50 values
for the virtual orbital selection density are shown here):
Eigenvalues and corresponding R_core values for the core orbital selection density
Orbital Eigenvalue R_core

0 0.00026936 0.419318
1 0.00027080 0.840883
2 0.00038739 1.443947
3 0.00038739 2.047011
4 0.00040299 2.674355
5 0.00040299 3.301700
6 0.00077636 4.510285
7 0.00086085 5.850394
8 0.00086085 7.190503
9 0.00091850 8.620358
10 0.00091850 10.050213
11 0.00112826 11.806598
12 0.00115561 13.605563
13 0.00137961 15.753236
14 0.00137961 17.900908
15 0.00139093 20.066210
16 0.00139093 22.231512
17 0.00143349 24.463072
18 0.00143350 26.694633
19 0.00148539 29.006985
20 0.00148539 31.319338
21 0.00173415 34.018940
22 0.00224131 37.508054
23 0.00224132 40.997171
24 0.00533017 49.294785
25 0.00533019 57.592429
26 0.00658679 67.846267
27 0.00662033 78.152314
28 0.00701718 89.076149
29 0.00701719 100.000000
Eigenvalues and corresponding R_virt values for the virtual orbital selection density
Orbital Eigenvalue R_virt

0 0.00000119 0.000450
1 0.00000119 0.000899
2 0.00000134 0.001404
3 0.00000134 0.001909
4 0.00000136 0.002423
5 0.00000177 0.003091
6 0.00000178 0.003764
7 0.00000178 0.004437
8 0.00000215 0.005248
9 0.00000224 0.006096
10 0.00000224 0.006944
11 0.00000238 0.007844
12 0.00000347 0.009154
13 0.00000347 0.010465
14 0.00000364 0.011841
15 0.00000386 0.013299
16 0.00000396 0.014793
17 0.00000396 0.016287
18 0.00000437 0.017937
19 0.00000437 0.019587
20 0.00000499 0.021472
21 0.00000499 0.023357
22 0.00000794 0.026354
23 0.00000794 0.029352
24 0.00000819 0.032447
25 0.00000819 0.035543
26 0.00000927 0.039044
27 0.00000927 0.042546
28 0.00001002 0.046332
29 0.00001002 0.050119
30 0.00001137 0.054415
31 0.00001137 0.058711
32 0.00001158 0.063086
33 0.00001158 0.067461
34 0.00001381 0.072678
35 0.00001381 0.077894
36 0.00001417 0.083249
37 0.00001417 0.088604
38 0.00001465 0.094137
39 0.00001495 0.099785
40 0.00001495 0.105432
41 0.00001554 0.111302
42 0.00001554 0.117172
43 0.00001623 0.123303
44 0.00001689 0.129685
45 0.00001754 0.136310
46 0.00001754 0.142934
47 0.00001805 0.149752
48 0.00001805 0.156570
49 0.00002111 0.164546
.
.
.
In conclusion, the orbital selection scheme provides a more efficient way to calculate accurate excitation
spectra within the framework of MR-EOM. It can be used to extend the applicability of this approach to
larger systems and we expect it to be much more effective in larger systems where the chromophore is localized
to a small part of the molecule. We reiterate that it is currently limited to the calculation of excitation
energies and should not be used if one is interested in total energies.
9.24.5 Nearly Size Consistent Results with MR-EOM by Employing an MR-CEPA(0)

Shift in the Final Diagonalization Procedure
One drawback of the MR-EOM methodology is that it is not size-extensive (or size-consistent). The size-
extensivity errors arise due to the final uncontracted MR-CI diagonalization step. Namely, they result
from the components of the eigenvectors of the transformed Hamiltonian, which lie outside of the CASSCF
reference space (e.g. 1h, 1p, etc. configurations). As more of the excitation classes are included through
the successive similarity transformations of the Hamiltonian, the size of the final diagonalization manifold is
greatly decreased resulting in much smaller size-extensivity errors upon going from MR-EOM-T|T -h-v to
MR-EOM-T|T |SXD|U-h-v. To illustrate this, let us consider the O2 O2 dimer where the O2 molecules are
separated by a large distance. For the O2 monomer, we employ a minimal active space consisting of 2 electrons
distributed amongst the two orbitals and we only consider the ground 3 g state (no state-averaging). In
1 1 +
the MR-EOM calculations, we also calculate the higher lying g and g singlet states. For example, the
input file for the MR-EOM-T|T |SXD|U-h-v calculation is given by:
!MR-EOM AUG-CC-PVTZ EXTREMESCF
%casscf
nel 2
norb 2
nroots 1
mult 3
end
%mdci
MaxIter 300
STol 1e-12
end
%mrci
newblock 1 *
nroots 3
refs cas(2,2) end
end
newblock 3 *
nroots 1
refs cas(2,2) end
end
end
* xyz 0 3
O 0.00000000 -0.00000000 -0.60500000
O -0.00000000 0.00000000 0.60500000
*
In the case of the dimer, we take the reference state as the coupled quintet state which is formed as the
product 3 + 3 +
g g of the monomer states. We note that at large separation, in the non-interacting limit, the
dimer state energies can be decomposed as the sum of monomer state energies. There are various possibilities,
taking into account the degeneracies of the various states:
1. a singlet, a triplet and a quintet with energy E(3 3

g + g ),
2. four triplets with energy E(3 1

g + g ),
3. two triplets with energy E(3 1 +

g + g ),
4. four singlets with energy E(1 g + 1 g ),
5. four singlets with energy E(1 g + 1 +

g ),
6. a singlet with energy E(1 + 1 +

g + g ).
Hence, in the final diagonalization step of the MR-EOM calculation, we must ask for 10 singlets, 7 triplets
and 1 quintet. The input file for the MR-EOM-T|T |SXD|U-h-v calculation on the dimer is given by:
!MR-EOM AUG-CC-PVTZ EXTREMESCF
%casscf
nel 4
norb 4
nroots 1
mult 5
etol 1e-13
gtol 1e-13
end
%mdci
MaxIter 300
STol 1e-12
end
%mrci
newblock 1 *
nroots 10
refs cas(4,4) end
end
newblock 3 *
nroots 7
refs cas(4,4) end
end
newblock 5 *
nroots 1
refs cas(4,4) end
end
end
* xyz 0 5
O 0.00000000 0.00000000 -500.60500000
O 0.00000000 -0.00000000 -499.39500000
O -0.60500000 0.00000000 500.00000000
O 0.60500000 -0.00000000 500.00000000
*
In Table 9.17, we have compiled the results of the size consistency test, taking the difference of the dimer
state energies (at large separation) and the sum of the monomer state energies (in mEh ). It is evident
that as more excitation classes are included in the similarity transformed Hamiltonian and the size of the
final diagonalization manifold is decreased, the size-consistency errors decrease. Of particular note are the
results for the MR-EOM-T|T |SXD|U-h-v approach (only includes 1h and 1p configurations in the final
diagonalization manifold), for which the largest deviation is 1.25 102 mEh . The much larger deviations
for the MR-EOM-T|T |SXD-h-v approach clearly demonstrate the large effect that the 2h excitations have
on the size-consistency errors.
Table 9.17: Test for size consistency in MR-EOM: Differences in energy (in mEh ) between the O2
O2 dimer energies (at large separation) and the sum of the monomer energies for the
ground state and various excited states. The results were obtained in an aug-cc-pVTZ
basis using minimal active spaces.
T|T -h-v T|T |SXD-h-v (with 1h1p) T|T |SXD-h-v T|T |SXD|U-h-v
E(3 3
g + g ) 12.74 2.77 1.11 1.00 105
E(3 1
g + g ) 14.20 3.84 1.85 1.54 104
E(3 1 +
g + g ) 17.21 5.52 2.83 4.13 104
E(1 g + 1 g ) 15.69 3.10 2.31 5.26 103
E(1 g + 1 +
g) 18.83 7.52 4.76 5.89 103
E(1 + 1 +
g + g ) 22.34 10.75 7.31 1.25 102
In order to reduce the size-consistency errors, one can make use of the MR-CEPA(0) shift in the final
diagonalization step. This MR-CEPA(0) shift can easily be activated by adding the line
citype mrcepa_0
to the beginning of the %mrci block. The results of the size-consistency test with the use of the MR-CEPA(0)
shift are tabulated in Table 9.18. For each of the methods, we see a marked improvement over the results
of Table 9.17, which do not make use of the MR-CEPA(0) shift. The greatest improvement occurs in the
MR-EOM-T|T |SXD-h-v and the MR-EOM-T|T |SXD|U-h-v results. Namely, the errors in the former
case are on the order of nano Hartrees, while the errors in the MR-EOM-T|T |SXD|U-h-v results are not
detectable (sub nano Hartree), as the energy is only printed with nine decimal places. It is interesting to note
that upon adding the 1h1p configurations to the diagonalization manifold in the MR-EOM-T|T |SXD-h-v
calculations (i.e. with 1h1p), the size-consistency errors increase greatly. Hence, it appears that the use of
the MR-CEPA(0) shift is most effective at reducing the size-consistency errors resulting from the presence
of the 1h, 1p and 2h configurations in the final diagonalization manifold. In any case, one can easily take
advantage of this approach to obtain nearly size-consistent results with both the MR-EOM-T|T |SXD-h-v
and MR-EOM-T|T |SXD|U-h-v methods.
Table 9.18: Test for size consistency in MR-EOM, using the MR-CEPA(0) shift: Differences in
energy (in mEh ) between the O2 O2 dimer energies (at large separation) and the
sum of the monomer energies for the ground state and various excited states. The
results were obtained in an aug-cc-pVTZ basis using minimal active spaces and the
MR-CEPA(0) shift was applied in the final diagonalization in each case.
.

T|T -h-v T|T |SXD-h-v (with 1h1p) T|T |SXD-h-v T|T |SXD|U-h-v

E(3 3
g + g ) 2.75 103 0.01 2.00 106 0.00
E(3 1
g + g ) 0.06 0.07 0.00 0.00
E(3 1 +
g + g ) 0.14 0.15 4.00 106 0.00
E(1 g + 1 g ) 0.21 0.22 1.00 106 0.00
E(1 g + 1 +
g) 0.42 0.44 5.00 106 0.00
E( g + +
1 + 1
g) 0.82 0.87 9.00 106 0.00
9.25 Simulation and Fit of Vibronic Structure in Electronic Spectra,

Resonance Raman Excitation Profiles and Spectra with the
orca asa Program
In this section various aspects of the simulation and fit of optical spectra, including absorption, fluorescence,
and resonance Raman are considered. This part of the ORCA is fairly autonomous and can also be used in a
data analysis context, not only in a quantum chemistry mode. The program is called orca asa, where
ASA stands for Advanced Spectral Analysis. The program was entirely designed by Dr. Taras Petrenko.
The general philosophy is as follows: An ORCA run produces the necessary data to be fed into the orca asa
program and writes an initial input file. This input file may be used to directly run orca asa in order to
predict an absorption, fluorescence or resonance Raman spectrum. Alternatively, the input file may be edited
to change the parameters used in the simulations. Last but certainly not least the orca asa program can
be used to perform a fit of the model parameters relative to experimental data.
All examples below are taken from the paper:
Petrenko, T.; Neese, F. (2007) Analysis and Prediction of Absorption Bandshapes, Fluorescence
Bandshapes, Resonance Raman Intensities and Excitation Profiles using the Time Dependent
Theory of Electronic Spectroscopy. J. Chem. Phys., 127, 164319
Which must be cited if you perform any work with the orca asa program!
9.25 The orca asa Program 619
9.25.1 General Description of the Program
The program input comprises the following information: (1) model and specification of the model parameters
characterizing the electronic structure of a molecule, as well as lineshape factors; (2) spectral ranges and
resolution for simulations; (3) specification of vibrational transitions for rR excitation profile and spectra
generation; (4) certain algorithm-selecting options depending on the model; (5) fitting options.
All optional parameters (1)-(3) are given in the %sim block, and fitting options are in the %fit block. The
model parameters are specified within various blocks that will be described below. The program orca asa is
interfaced to ORCA and inherits its input style. The input for orca asa run can be also generated upon
ORCA run.
The current implementation features so called simple, independent mode, displaced harmonic oscillator
(IMDHO), and independent mode, displaced harmonic oscillator with frequency alteration (IMDHOFA)
models.
9.25.2 Spectral Simulation Procedures: Input Structure and Model Parameters
9.25.2.1 Example: Simple Mode
This model represents the simplest approach which is conventionally used in analysis of absorption spectra.
It neglects vibrational structure of electronic transitions and approximates each individual electronic band
by a standard lineshape, typically a Gaussian, Lorentzian or mixed (Voigt) function. This model can only
make sense if vibrational progressions are not resolved in electronic spectra. Upon this approximation
the intensity of absorption spectrum depends on the energy of the incident photon (EL ), the electronic
transition energy (ET ), the transition electric dipole moment (M, evaluated at the ground-state equilibrium
geometry). Lineshape factors are specified by homogeneous linewidth and standard deviation parameter
corresponding to Gaussian distribution of transition energies. The following example illustrates a simple
input for simulation of absorption bandshapes using various intensity and lineshape parameters.
# example001.inp
#
# Input file to generate absorption spectrum consisting
# of 3 bands with different lineshape factors:
#
# 1. Lorentzian centered at 18000cm**-1 (damping factor Gamma= 100 cm**-1)
# 2. Gaussian centered at 20000cm**-1
# (standard deviation Sigma= 100 cm**-1)
# 3. Mixed Gaussian-Lorentzian band representing Voigt profile
# centered at 21000 cm**-1
%sim
Model Simple
# Spectral range for absorption simulation:

AbsRange 17000.0, 23000.0
# Number of points to simulate absorption spectrum:

NAbsPoints 2000
end
#---------------------------------------------------------------------------
# Transition Gamma Sigma Transition Dipole Moment (atomic unit)
# Energy (cm**-1) (cm**-1) (cm**-1) Mx My Mz
#---------------------------------------------------------------------------
$el_states
3 # number of electronic states
1 18000.0 100.00 0.0 1.0 0.0 0.0
2 20000.0 0.00 100.0 1.0 0.0 0.0
3 22000.0 50.00 50.0 1.0 0.0 0.0
The parameters of of the final electronic states reached by the respective transitions are specified in the
$el states block. The spectral range and resolution used in the calculation are defined by the AbsRange
and NAbsPoints keywords in %sim block. The calculation of the absorption spectrum is automatically
invoked if NAbsPoints>1. After the orca asa run you will find in your directory file example001.abs.dat
containing absorption spectrum in simple two-column ASCII format suitable to be plotted with any spread-
sheet program. Absorption spectra corresponding to individual electronic transitions are stored in file
example001.abs.as.dat ( the suffix as stands for All States).
The output of the program run also contains information about oscillator strengths and full-width-half-
maximum (FWHM) parameters corresponding to each electronic band:
----------------------------------------------
State EV fosc Stokes shift
(cm**-1) (cm**-1)
----------------------------------------------
1: 18000.00 0.054676 0.00
2: 20000.00 0.060751 0.00
3: 22000.00 0.066826 0.00
Figure 9.29: Absorption spectrum generated after orca asa run on file example001.inp. Three
bands have different lineshape pararameters. Note that although all transitions are
characterized by the same transition electric dipole moment their intensities are scaled
proportionally to the transition energies.
----------------------------------------
----------------------------------------
State Gamma Sigma FWHM
----------------------------------------
1: 100.00 0.00 200.00
2: 0.00 100.00 235.48
3: 50.00 50.00 180.07
Note that although all three types of lineshape functions are symmetric this is not true for the overall shapes
of individual absorption bands since the extinction coefficient (absorption cross-section) is also proportional to
the incident photon energy. Therefore, if the linewidth is larger than 10% of the peak energy the asymmetry
of the electronic band can be quite noticeable.
9.25.2.2 Example: Modelling of Absorption and Fluorescence Spectra within the IMDHO
Model
The IMDHO model is the simplest approach that successfully allows for the prediction of vibrational structure
in electronic spectra as well as rR intensities for a large variety of real systems. This model assumes:
1. harmonic ground- and excited-state potential energy surfaces;
2. origin shift of the excited-state potential energy surface relative to the ground-state one;
3. no vibrational frequency alteration or normal mode rotation occurs in the excited state;
4. no coordinate dependence of the electronic transition dipole moment.

In addition to the parameters that enter the Simple model defined above it requires some information
about the vibrational degrees of freedom. The required information consists of the ground-state vibrational
frequencies {gm } and (dimensionless) origin shifts {mi }, where i and m refer to electronic states and
normal modes respectively. is expressed in terms of dimensionless normal coordinates. Accordingly, for the
IMDHO model one has to specify the following blocks
The $el states block contains the parameters ET , , , M for each electronic state. By default ET is
assumed to be adiabatic minima separation energy. Alternatively, it can be redefined to denote for the
vertical transition energy.This is achieved by specifiying the keyword EnInput=EV in the %sim block.
A $vib freq gs block specifies ground-state vibrational frequencies.
A $sdnc block contains parameters {mi } in matrix form such that the i-th column represents the
dimensionless displacements along all normal modes for the i-th excited-state PES.
The file example002.inp provides the input for simulation of absorption and fluorescence spectra of a system
characterized by significant displacements of the excited-state origin along 5 normal coordinates.
# example002.inp
#
# Input file for simulation of vibrational structure
# in absorption and fluorescence spectra assuming
# origin shift of excited PES along 5 normal coordinates.
# The simulated spectra closely reproduce the experimental
# optical bandshapes for the tetracene molecule.
#
%sim
Model IMDHO
# spectral range for absorption simulation (cm**-1)

AbsRange 20000.0, 27000.0
NAbsPoints 2000 # number of points in absorption spectrum
# spectral range for simulation of fluorescence (cm**-1)

FlRange 22000.0, 16000.0
NFlPoints 2000 # number of points in fluorescence spectrum
# the following options require the spectra to be normalized

# so that their maxima are equal to 1.0
AbsScaleMode Rel
FlScaleMode Rel # default for fluorescence
# for absorption spectrum the default option is AbsScaleMode= Ext

# which stands for extinction coefficient
end
#---------------------------------------------------------------------------
#---------------------------------------------------------------------------
$el_states
1
1 21140.0 50.00 100.0 1.0 0.0 0.0
# Block specifying Stokes Shift parameter for each electronic state

# This information is optional
$ss
1 # number of excited states
1 300.0 # the Stokes shift for the 1st electronic transition
# Block providing the values of VIBrational FREQuencies

# for 5 Ground-State normal modes.
# Obligatory for IMDHO and IMDHOFA models.
$vib_freq_gs
5
1 310.0
2 1193.0
3 1386.0
4 1500.0
5 1530.0
# Block specifying origin Shift of the excite-state PES

# along each normal mode in terms of the ground-state
# Dimensionless Normal Coordinates
# Obligatory for IMDHO and IMDHOFA models.
$sdnc
5 1
1
1 0.698
2 -0.574
3 0.932
4 -0.692
5 0.561
The calculation of absorption and fluorescence spectra is automatically invoked if the parameters NAbsPoints>1
and NFlPoints>1. The input file also contains the optional block $ss which specifies the Stokes shift
for each electronic transition. This parameter is equal to the energy separation between the 0-0 vibrational
peaks in the absorption and fluorescence spectra as shown in Figure 9.30 . In general accounts for solvent
induced effects as well as unresolved vibrational structure corresponding to low-frequency modes that are not
specified in the input. Note that we have specified parameters AbsScaleMode=Rel and FlScaleMode=Rel in
%sim block in order to ensure that the simulated spectra are normalized to unity. The calculated absorption
and fluorescence spectra are stored in example002.abs.dat and example002.fl.dat files, respectively.
9.25.2.3 Example: Modelling of Absorption and Fluorescence Spectra within the IMDHOFA
Model
IMDHOFA (Independent Mode Displaced Harmonic Oscillators with Frequency Alteration) is based on the
same assumptions as the IMDHO model except for vibrational frequency alteration in excited state can
take place. The file example003.inp features almost the same input parameters as example002.inp. The
IMDHOFA model is invoked by the keyword Model=IMDHOFA in the %sim block. Additionally, one has to
provide the obligatory block $vib freq es. It contains the excited-state vibrational frequencies {emi } in
Figure 9.30: Absorption and fluorescence spectra generated after orca asa run on the file
example002.inp. If the homogeneous broadening is set to be = 10 cm1 one can
resolve underlying vibrational structure and identify various fundamental and combi-
nation transitions.
matrix form such that the i-th column represents the vibrational frequencies of all normal modes for the i-th
excited-state PES.

# for 5 Excited-State normal modes.
# Obligatory for IMDHOFA model.
$vib_freq_es
5 1 # number of modes and number of excited states
1
1 410.0
2 1293.0
3 1400.0
4 1600.0
5 1730.0
Figure 9.31: Absorption and fluorescence spectra generated after orca asa run on the file
example003.inp. Also, the high-resolution spectra corresponding to homogeneous
broadening = 10 cm1 are shown.
9.25.2.4 Example: Modelling of Effective Broadening, Effective Stokes Shift and

Temperature Effects in Absorption and Fluorescence Spectra within the IMDHO
Model
For the IMDHO model the orca asa is capable to model absorption and emission spectra in the finite-
temperature approximation. While the keyword Model=IMDHO assumes the zero-temperature approximation,
the value of Model=IMDHOT invokes the calculation of the spectra for the finite temperature which is specified
by the paramter TK in the block %sim:
# example004.inp
#
#
%sim
Model IMDHOT
TK 300 # temperature (in Kelvin)
# spectral range for absorption simulation (cm**-1)

AbsRange 18000.0, 35000.0
NAbsPoints 5000 # number of points in absorption spectrum
# spectral range for simulation of fluorescence (cm**-1)

FlRange 22000.0, 10000.0
NFlPoints 5000 # number of points in fluorescence spectrum
# the following options require the spectra to be normalized

# so that their maxima are equal to 1.0
AbsScaleMode Rel
FlScaleMode Rel # default for fluorescence
end
#---------------------------------------------------------------------------
#---------------------------------------------------------------------------
$el_states
1
1 21140.0 50.00 100.0 1.0 0.0 0.0
# Block specifying Stokes Shift parameter for each electronic state

$ss
1 # number of excited states
1 300.0 # the Stokes shift for the 1st electronic transition

# for 10 Ground-State normal modes.
$vib_freq_gs
10
1 30.0
2 80.0
3 100.0
4 120.0
5 130.0
6 140.0
7 160.0
8 200.0
9 310.0
10 1300.0
# Block specifying origin Shift of the excite-state PES

# along each normal mode in terms of the ground-state
# Dimensionless Normal Coordinates
$sdnc
10 1
1
1 2.5
2 2.0
3 1.8
4 1.9
5 1.5
6 1.9
7 2.4
8 1.9
9 2.5
10 0.9
This example illustrates a typical situation in large molecules which feature a number of low frequency modes
with significant values of dimensionless displacements for a given excited-state PES. In the case of high density
of vibrational states with frequencies below or comparable to the intrincic value of FWHM (determined
by and ) the vibrational progression is unresolved, whereby the spectra become very diffuse and show
large separation between the maxima of absorption and emission spectra (Figure 9.31). Besides, upon the
condition hi 6 kT the effective bandwidths and positions of maxima in the spectra can be strongly subject
to temperature effects.
Figure 9.32: Absorption and fluorescence spectra for T=0 K (blue) and T=300 K (red) generated
after orca asa run on the file example004.inp. Black lines show spectra corresponding
to the case where all low-frequency modes were excluded from the calculation.
The effective Stokes shift and linewidth parameters which are evaluated in the simple self-consistent procedure
are given in the output of the orca asa run:
------------------------------------------------------------------------------
State E0 EV fosc Stokes shift Effective Stokes shift
(cm**-1) (cm**-1) (cm**-1) (cm**-1)
------------------------------------------------------------------------------
1: 21140.00 24535.85 0.074529 300.00 7091.70
-----------------------------------------------------------------------------------------------
-----------------------------------------------------------------------------------------------
Intrinsic Effective
State -------------------------- --------------------------------------------------------
Sigma FWHM
Gamma Sigma FWHM --------------------------- ---------------------------
0K 298.15K 300.00K 0K 298.15K 300.00K
-----------------------------------------------------------------------------------------------
1: 50.00 100.00 293.50 1125.34 1411.13 1413.57 2703.84 3376.75 3382.48
Note that the evaluation of the effective parameters is rather approximate and these values can noticeable
deviate from those which can be directly deduced from the calculated spectra. However, such an information
usually provides the proper order of magnitude of the effective vibronic broadening and Stokes shift. As
indicated in the program output above, the effective bandshape has predominantly a Gaussian character
which varies with the temperature so that = 1125 cm1 (T = 0 K) and = 1414 cm1 (T = 300 K). Indeed,
as shown in Figure 9.33 the absorption spectrum at T = 300 K can be well fitted using Gaussian lineshape
with = 1388 cm1 (FWHM= 3270 cm1 ). One can see that at higher temperatures the deviation between
the spectrum and its Gauss fit becomes even smaller.
In molecules the normal distribution of the electronic transition energies in the ensemble would give rise to a
Gaussian bandshape of the absorption band. However, the corresponding standard deviation is expected to
be of the order of 100 cm1 , whereby a typical Gaussian bandwidth of the order of 1000 cm1 appears to
result from unresolved vibronic progression. In general, this statement is supported by quantum chemical
calculation of the model parameters. In principle the effective bandwidth parameters can also be used for
characterization and assignement of individual electronic bands.
Figure 9.33: Absorption spectrum (blue) for T = 300 K generated after orca asa run on the file
example004.inp. Red line represents the Gauss-fit of the calculated spectrum.
9.25.2.5 Example: Modelling of Absorption and Resonance Raman Spectra for the 1-1 Ag
1-1 Bu Transition in trans-1,3,5-Hexatriene
The hexatriene molecule is characterized by 9 totally-symmetric normal modes which dominate vibrational
structure in absorption and are active in rR spectra corresponding to the strongly dipole-allowed 1 1 Ag
1 1 Bu transition around 40000 cm1 . Except for some peculiarities related to the neglect of normal mode
rotations in the excited state the optical spectra are quite satisfactorily described by the IMDHO model.
The following input exemplifies simulation of absorption spectrum and rR spectra for an arbitrary predefined
number of excitation energies.
#
# example005.inp
#
# input for simulation of absorption and resonance Raman spectra
# using experimental values of transition energy and displacement
# parameters corresponding to the strongly allowed 1-1Ag 1-1Bu transition
# in trans-1,3,5-hexatriene
#
%sim
Model IMDHO
AbsRange 38000.0, 48000.0

NAbsPoints 2000
AbsScaleMode Rel
# resonance Raman intensities will be calculated

# for all vibrational states with excitation number
# up to RamanOrder:
RamanOrder 4
# excitation energies (cm**-1) for which rR spectra will be calculated:

RRSE 39500, 39800, 41400
# full width half maximum of Raman bands in rR spectra (cm**-1):

RRS_FWHM 10
RSRange 0, 5000 # spectral range for simulation of rR spectra (cm**-1)

NRRSPoints 5000 # number of points to simulate rR spectra (cm**-1)
end
$el_states
1
1 39800.0 150.00 0.0 1.0 0.0 0.0
$vib_freq_gs
9
1 354.0
2 444.0
3 934.0
4 1192.0
5 1290.0
6 1305.0
7 1403.0
8 1581.0
9 1635.0
$sdnc
9 1
1
1 0.55
2 0.23
3 0.23
4 0.82
5 0.485
6 0.00
7 0.085
8 0.38
9 1.32
After the orca asa run the following files will be created:
example005abs.dat contains the simulated absorption spectrum. It is shown in Figure 9.34.
example005.o4.rrs.39500.dat, example005.o4.rrs.39800.dat and example005.o4.rrs.41400.dat

contain the simulated rR spectra for excitation energies at 39500, 39800 and 41400 cm1 , respec-
tively. The suffix o4 stands for the order of Raman scattering specified in the input by keyword
RamanOrder=4. The rR specta are shown in Figure 9.35.
example005.o4.rrs.39500.stk, example005.o4.rrs.39800.stk and example005.o4.rrs.41400.stk

provide Raman shifts and intensities for each vibrational transition. Corresponding vibrational states
are specified by the quantum numbers of excited modes.
Figure 9.34: Absorption spectrum corresponding to 1 1 Ag 1 1 Bu transition in trans-1,3,5-

hexatriene generated after orca asa run on the file example005.inp.
Figure 9.35: Resonance Raman spectra for 3 different excitation energies which fall in resonance
with 1 1 Ag 1 1 Bu transition in trans-1,3,5-hexatriene.
NOTE
By default the program provides rR spectra on an arbitrary scale since only relative rR intensities
within a single rR spectrum are of major concern in most practical cases. However, one can put rR
spectra corresponding to different excitation energies on the same intensity scale by providing the
keyword RSISM=ASR in %sim block (RSISM Raman Spectra Intensity Scaling Mode; ASR All
Spectra Relative). By default RSISM=SSR (SSR Single Spectrum Relative) for which each rR spectrum
is normalized so that the most intense band in it has intensity 1.0. The relative intensities of bands
in rR spectra measured for different excitation energies can be compared if they are appropriately
normalized relative to the intensity of a reference signal (e.g. Raman band of the solvent). We also
keep in mind the possibility to extend our methodology in order to provide the absolute measure of rR
intensities in terms of the full or differential cross-sections.
Within the harmonic model, for a single electronic state neither relative rR intensities nor absorption
bandshapes in the case of AbsScaleMode=Rel do depend on the values of the electronic transition
dipole moment (unless it is precisely zero).
In the example above resonance Raman spectra have been generated for all vibrational transitions with
total excitation number up to the value specified by the parameter RamanOrder. Its is also possible to make
explicit specification of vibrational states corresponding to various fundamental, overtone and combination
bands via the $rr vib states block. In such a case rR spectra involving only these vibrational transitions
will be generated separately.
$rr_vib_states 5 # total number of vibrational transitions

1
modes 1
quanta 1; # final vibrational state for the fundamental band corresponding to mode 1
2
modes 9
quanta 1; # final vibrational state for the fundamental band corresponding to mode 9
3
modes 3, 4
quanta 1, 1; # final vibrational state for the combination band involving single
# excitations in modes 3 and 4
4
modes 5
quanta 3; # final vibrational state for the second overtone band corresponding to
# mode 5
5
modes 1, 5,9
quanta 1,2, 1; # final vibrational state for the combination band involving single
# excitations in modes 1 and 2, and double excitation in mode 5
Each vibrational transition is specified via the subblock which has the following structure:
k
modes m1,m2,...mn
quanta q1,q2,...qn;
This means that the k-th transition is characteriezed by excitation numbers qi for modes mi so that
P
corresponding Raman shift is equal to = qi i , where i is vibrational frequency of the mode mi .
After the orca asa run the following files will be created in addition:
example005.us.rrs.39500.dat, example005.us.rrs.39800.dat and example005.us.rrs.41400.dat

contain the simulated rR spectra involving only vibrational transitions specified in the $rr vib states
block, for excitations energies at 39500, 39800 and 41400 cm1 , respectively. The suffix us stands for
User specified vibrational States.
example005.us.rrs.39500.stk, example005.us.rrs.39800.stk and example005.us.rrs.41400.stk

provide Raman shifts and intensities for each vibrational transition specified in the $rr vib states
block.
9.25.2.6 Example: Modelling of Absorption Spectrum and Resonance Raman Profiles for the
1-1 Ag 1-1 Bu Transition in trans-1,3,5-Hexatriene
The following example illustrates an input for simulation of absorption bandshape and resonance Raman
profiles (RRP):
#
# example006.inp
#
# input for simulation of absorption and resonance Raman profiles
# using experimental values of transition energy and displacement
# parameters corresponding to the strongly allowed 1-1Ag 1-1Bu transition
# in trans-1,3,5-hexatriene
#
%sim
Model IMDHO
AbsRange 38000.0, 48000.0

NAbsPoints 2000
AbsScaleMode Rel
RRPRange 38000.0, 48000.0 # spectral range for simulation of

# rR profiles (cm**-1)
NRRPPoints 2000 # number of points for simulation of rR profiles

CAR 0.8
RamanOrder 2
end
$el_states
1
1 39800.0 150.00 0.0 1.0 0.0 0.0
$vib_freq_gs
9
1 354.000000
2 444.000000
3 934.000000
4 1192.000000
5 1290.000000
6 1305.000000
7 1403.000000
8 1581.000000
9 1635.000000
$sdnc
9 1
1
1 0.55
2 0.23
3 0.23
4 0.82
5 0.485
6 0.00
7 0.085
8 0.38
9 1.32
$rr_vib_states 5 # total number of vibrational transitions

1
modes 1
quanta 1;
2
modes 9
quanta 1;
3
modes 3, 4
quanta 1, 1;
4
modes 5
quanta 3;
5
modes 1, 5,9
quanta 1,2, 1;
The keyword RamanOrder=2 will invoke generation of rR profiles for all vibrational transitions with total
excitation number up to 2 in the range of excitation energies specified by the keywords RRPRange and
NRRPPoints. Likewise, rR profiles for the vibrational states given in the $rr vib states block will be
generated separately. Since in most cases only relative rR intensities are important, and one would be
interested to compare absorption bandshape and shapes of individual rR profiles, the keyword CAR = 0.8 is
used to scale rR profiles for all vibrational transitions by a common factor in such a way that the ratio of the
maximum of all rR intensities and the maximum of absorption band is equal to 0.8.
After the orca asa run the following files will be created:
example006.abs.dat contains the simulated absorption spectrum (Figure 9.36).
example006.o1.rrp.dat and example006.o2.rrp.dat contain rR profiles for vibrational transitions

with total excitation numbers 1 and 2, respectively. RR profiles for all fundamental bands (from the
file example006.o1.rrp.dat) are shown in Figure 9.36.
example006.o1.info and example006.o1.info contain specification of vibrational transitions with

total excitation numbers 1 and 2, respectively, as well as corresponding Raman shifts.
example006.us.rrp.1.dat--example006.us.rrp.5.dat contain rR profiles for vibrational transitions

15 specified in the $rr vib states block.
Figure 9.36: Absorption spectrum and resonance Raman profiles of fundamental bands correspond-
ing to 1 1 Ag 1 1 Bu transition in trans-1,3,5-hexatriene.
9.25.3 Fitting of Experimental Spectra
9.25.3.1 Example: Gauss-Fit of Absorption Spectrum
An absorption spectrum basically consists of a number of absorption bands. Each absorption band corresponds
to a transition of the ground electronic state to an excited electronic state. In molecules such transitions
are usually considerably broadened. In many cases there will be overlapping bands and one would need
to deconvolute the broad absorption envelope into contributions from individual transitions. Within the
Simple model the orca asa program enables fit of an absorption spectrum with a sum of standard lineshape
functions (Gaussian, Lorentzian) or more general Voigt functions. In most cases, one simply performs a
Gauss-Fit. That is, it is assumed that the shape of each individual band is that of a Gaussian function. Then
one applies as many (or as few) Gaussians as are necessary for an accurate representation of the absorption
envelope. In order to explain the fitting procedures within the Simple model let us consider an experimental
absorption spectrum in Figure 9.37:
Figure 9.37: Experimental absorption spectrum. Bars indicate transition energies which were used
for the initial guess in the input for spectral fitting.
As shown in Figure 9.37 one can identify roughly 7 electronic bands. The initial estimates of transition energies
corresponding to the maxima and shoulders in the absorption spectrum (indicated by bars in Figure 9.37)
and rather approximate values of inhomogeneous broadening and transition dipole moment components are
specified in the $el states block of the input file for the spectral fitting:
# example007.inp
#
# Input file for fitting of experimental absorption spectrum
#
%sim
model Simple
end
%fit
Fit true # Global flag to turn on the fit
AbsFit true # Flag to include absorption into the fit
method Simplex
WeightsAdjust true
AbsRange 0.0, 100000.0 # absorption spectral range to be included in the fit;

# in the present case all experimental points
# will be included
AbsName "absexp.dat" # name of the file containing experimental

# absorption spectrum in a simple two-column
# ASCII format
ExpAbsScaleMode Ext # This keyword indicates that the experimental

# absorption intensity is given in terms of

# the extinction coefficient. This is important
# for the proper fitting of transition dipole
# moments and oscillator strengths
NMaxFunc 10000 # maximum number of function evaluations in simplex

# algorithm
MWADRelTol 1e-5 # Relative Tolerance of the Mean Weighted Absolute

# Difference (MWAD) function which specifies the
# convergence criterion
E0Step 500.00 # initial step for the transition energies

# in the simplex fitting
TMStep 0.5 # initial step for the transition dipole moments

E0SDStep 500.0 # initial step for the inhomogeneous linewidth (Sigma)

end
# ! Parameters specified in the $el_states block

# are used as initial guess in the fit
#---------------------------------------------------------------------------
#---------------------------------------------------------------------------
$el_states
7
1 11270 0.0 1000.00 1.0000 0.0000 0.0000
2 15100 0.0 1000.00 1.0000 0.0000 0.0000
3 20230 0.0 1000.00 1.0000 0.0000 0.0000
4 27500 0.0 1000.00 1.0000 0.0000 0.0000
5 31550 0.0 1000.00 1.0000 0.0000 0.0000
6 37070 0.0 1000.00 1.0000 0.0000 0.0000
7 39800 0.0 1000.00 1.0000 0.0000 0.0000
# the integer values specified in $el_states_c block indicate parameters

# in the $el_states block to be varied
$el_states_c
7
1 1 0 1 1 0 0
2 2 0 2 2 0 0
3 3 0 3 3 0 0
4 4 0 4 4 0 0
5 5 0 5 5 0 0
6 6 0 6 6 0 0
7 7 0 7 7 0 0
The functionality of the constraint block $el states c should be understood as follows: 1) 0 flag indicates
that the corresponding parameter in the $el state block will not be varied in the fitting; 2) if the number
corresponding to a certain parameter coincides with the number of the corresponding electronic state this
parameter will be varied independently. Thus, the block $el states c in the input indicates that all transition
energies, inhomogeneous linewidths and x-components of the transition electric dipole moment will be varied
independently, while homogeneous linewidths, y- and z-components of the transition dipole moment will be
fixed to their initial values.
The following considerations are important:
Since in conventional absorption spectroscopy one deals with the orientationally averaged absorption
cross-section, the signal intensity is proportional to the square of the transition electric dipole moment
2
|M| . Thus, the intensities do not depend on the values of the individual components of M as long
2
as |M| = const. Therefore, we have allowed to vary only Mx components. Otherwise there can be
problems in convergence of the fitting algorithm.
The sum of the weights of experimental points which enter the mean absolute difference function
employed in the the minimization is always kept equal to the number of experimental points. In the
case of equidistant experimental photon energies all weights are assumed to be equal. However, in
experimental electronic spectra the density of spectral points can increase significantly upon going from
high- to low-energy spectral regions, which is due to the fact that experimental absorption spectra are
initially acquired on the wavelength scale. In such a case the quality of the fit can be noticeably biased
towards low-energy spectral region. Therefore, it is advisable to adjust relative weights of experimental
points according to the their density which is controlled by the keyword WeightsAdjust in the %fit
block. Although this parameter is not crucial for the present example, in general, it will provide a
more balanced fit.
The parameters E0Step, TMStep, E0SDStep in the %fit block specify the initial dimension of the
simplex in the space of ET , M, and should roughly correspond to the expected uncertainty of initial
guess on these parameters in the $el states block relative to their actual values. The quality of the
fit can noticeably deteriorate if the parameters specifying initial steps are too low or too high.
The fit run of orca asa on file example007.inp will converge upon approximately 3600 function evaluations
(for MWADRelTol=1e-5). The results of the fit will be stored in file example007.001.inp which has the same
structure as the input file example007.inp. Thus, if the fit is not satisfactory and/or it is not fully converged
it can be refined in a subsequent orca asa run upon which file example007.002.inp will be created, and so
on. Some model parameters in intermediate files can be be additionally modified and/or some constraints can
be lifted or imposed if so desired. The output file example007.001.inp will contain fitted model parameters
stored in the $el states block:
$el_states
7
1 11368.24 0.00 732.50 1.6290 0.0000 0.0000
2 15262.33 0.00 495.17 -0.2815 0.0000 0.0000
3 19500.08 0.00 1023.39 0.2300 0.0000 0.0000
4 26969.01 0.00 1832.30 1.4089 0.0000 0.0000
5 31580.41 0.00 1440.87 1.8610 0.0000 0.0000
6 35769.07 0.00 1804.02 1.5525 0.0000 0.0000
7 39975.11 0.00 1909.38 2.4745 0.0000 0.0000
The overall quality of the fit is determined by the parameter MWAD which upon convergence reaches the
value of 0.009 (MWAD stands for Mean Weighted Absolute Difference).
After the orca asa run files absexp.fit.dat and absexp.fit.as.dat will be created. Both files contain the
experimental and fitted spectra which are shown in Figure 9.38 . In addition, the file absexp.fit.as.dat
will contain individual contributions to the absorption spectrum corresponding to different excited states.
Figure 9.38: Comparison of the experimental (black curve) and fitted (red) absorption spectra
corresponding to the fit run of orca asa on the file example007.inp. Blue curves
represent individual contributions to the absorption spectrum from each state.
Since there is a noticeable discrepancy between the fitted and experimental spectra around 13000 cm1 (Figure
9.38) it is worthwhile to refine the fit after adding parameters for a new state in the file example007.001.inp:
$el_states
8
1 11368.24 0.00 732.50 1.6290 0.0000 0.0000
... ... ...
8 13280.00 0.00 1000.00 1.000 0.0000 0.0000
$el_states_c
8
1 1 0 1 1 0 0
... ... ...
8 8 0 8 8 0 0
Actually, the character of the discrepancy in the present case is very similar to that in Figure 9.35 (section
9.25.2.4) where a vibronically broadened absorption spectrum was fitted with a Gaussian lineshape. Thus,
the poor fit in the region around 1300 cm1 is most likely due to the essentially asymmetric character of the
vibronic broadening rather than to the presence of another electronic band.
As shown in Figure 9.39 the refined fit leads to much better agreement between the experimental and fitted
absorption spectra (MWAD=0.0045).
Figure 9.39: Comparison of the experimental (black) and fitted (red) absorption spectra corre-
sponding to the fit run of orca asa on the file example007.001.inp. Blue curves
represent individual contributions to the absorption spectrum from each state.
Due to some peculiarities of the simplex algorithm for function minimization, you can still refine the fit by
rerunning orca asa on the file example007.002.inp! This leads to an even lower value of the parameter
MWAD= 0.0038, and therefore to better agreement of experimental and fitted spectra (even though the
previous run has been claimed to be converged).
It is also possible to perform a fit using the same value of inhomogeneous linewidth for all electronic states.
For this purpose one needs to choose as a guess the same linewidth parameters in the $el states block:
$el_states
8
1 11118.58 0.00 1000.0 1.0687 0.0000 0.0000
2 13673.38 0.00 1000.0 -0.5530 0.0000 0.0000
3 21267.40 0.00 1000.0 0.3675 0.0000 0.0000
4 27024.71 0.00 1000.0 1.4041 0.0000 0.0000
5 31414.74 0.00 1000.0 1.7279 0.0000 0.0000
6 35180.77 0.00 1000.0 1.6246 0.0000 0.0000
7 39985.52 0.00 1000.0 2.5708 0.0000 0.0000
8 11665.01 0.00 1000.0 1.2332 0.0000 0.0000
In addition the constraint block should be modified as follows:
$el_states_c
8
1 1 0 1 1 0 0
2 2 0 1 2 0 0
3 3 0 1 3 0 0
4 4 0 1 4 0 0
5 5 0 1 5 0 0
6 6 0 1 6 0 0
7 7 0 1 7 0 0
8 8 0 1 8 0 0
The constraint parameters for the inhomogeneous broadening were chosen to be 1, which means that formally
1 corresponding to the first state is varied independently while the linewidths {i } for other bands are
varied in such a way that the ratios i /1 are kept fixed to their initial values, whereby the same linewidth
parameter will be used for all states.
Figure 9.40: Comparison of the experimental (black) and fitted (red) absorption spectra corre-
sponding to the fit run of orca asa on the file example007.002.inp in which equal
broadening was assumed for all electronic bands. Blue curves represent individual
contributions to the absorption spectrum from each state.
One can see (Figure 9.40) that the assumption of equal linewidths for all electronic bands leads to a rather
pronounced deterioration of the quality of the fit in the low-energy spectral range (MWAD=0.017). Apparently,
this discrepancy can be fixed assuming more electronic states at higher energies.
NOTE
The homogeneous linewidth parameters can also be included in the fit in a similar way. However, one
can see that in most cases they appear to be much smaller than corresponding Gaussian linewidth
parameters.
Gauss-fit of absorption spectra is coventionally performed assuming the same linewidth parameters
for all bands. However, since a large portion of Gaussian broadening is mainly due to the unresolved
vibronic structure in the spectra which can significantly vary depending on the nature of transition,
the assumption of unequal Gaussian bandwidths seems to be a physical one.
9.25.3.2 Example: Fit of Absorption and Resonance Raman Spectra for 1-1 Ag 1-1 Bu
Transition in trans-1,3,5-Hexatriene
Below we provide an example of the fit of the lineshape parameters and {m } corresponding to the strongly
dipole-allowed 1-1 Ag 1-1 Bu transition in hexatriene. It is known that the most intense bands in rR spectra
correspond to the most vibronically active in absorption spectrum. For the IMDHO model this correlation
is determined by the values of {m }. Thus, the larger , the larger is the rR intensity of a given mode
and the more pronounced is the progression in the absorption spectrum corresponding to this mode. In
principle, if all vibrational transitions in absorption are well resolved it is possible to determine {m } by
a fit of the absorption spectrum alone. In practice this task is ambiguous due to the limited resolution of
the experimental absorption spectra. The observation of a rR spectrum enables the identification of the
vibrational modes that are responsible for the progression in the absorption spectrum, as well as a quantitative
analysis in terms of {m }. The file example006.inp provides a brute-force example on how to approach
the fit employing the minimal possible experimental information: 1) An absorption spectrum; 2) relative
rR intensities of fundamental bands for a given excitation energy. The rR spectrum upon the excitation in
resonance with the 0-0 vibronic band at 39809 cm1 is shown in Figure 9.29.
Figure 9.41: Experimental Resonance Raman spectrum corresponding to 1-1 Ag 1-1 Bu transition
in trans-1,3,5-hexatriene.
The experimental rR spectrum has enabled the identification of seven vibrational modes that give rise to
the most intense resonance Raman bands. Therefore, they are expected to have the largest excited-state
displacements and the most pronounced effect on the vibrational structure of the absorption spectrum. Their
vibrational frequencies have been entered as input for the fit as shown below:
#
# example008.inp
#
# Input for fit of absorption and resonance Raman spectra
# corresponding to the strongly allowed 1-1Ag 1-1Bu transition
# in 1,3,5 trans-hexatriene.
#
# Parameters to be varied:
# 1) adiabatic minima transiton energy
# 2) homogeneous linewidth (Gamma)
# 3) dimensionless normal coordinate displacements of the
# excited-state origin
#
%sim
Model IMDHO
end
%fit
Fit true # boolean parameter to switch on the fit
# boolean parameter to include experimental absorption

# spectrum in the fit:
AbsFit true
# boolean parameter to include experimental rR spectra

# specified in $rrs_exp block in the fit:
RRSFit true
AbsExpName "hex-abs.dat" # name of the file with experimental absorption

# spectrum
# the following value of keyword ExpAbsScaleMode

# indicates that only the shape of absorption band
# but not its total intensity will be accounted in the fit:
ExpAbsScaleMode Rel
# the weight of absorption relative to the total weight of

# rR intensities in the difference function to be minimized:
CWAR 5.0

# algorithm
MWADRelTol 1e-4 # Relative Tolerance of the Mean Weighted Absolute

SDNCStep 1.0
end
# The values specified in $el_states block serve as initial guess in the fit
$el_states
1
1 40000.0 200.00 0.0 1.0 0.0 0.0
# the integer values specified in $el_states_c block indicate parameters

# in $el_states block to be varied
$el_states_c
1
1 1 1 0 0 0 0
# 7 totally symmetric vibrations which give rise to the most

# intense bands in the rR spectra are included into analysis.
# Experimental values of vibration frequencies are given:
$vib_freq_gs
7
1 354.0
2 444.0
3 934.0
4 1192.0
5 1290.0
6 1403.0
7 1635.0
# Initial guess for the values of dimensionless normal

# coordinate displacements of the excited-state origin
$sdnc
7 1
1
1 0.0
2 0.0
3 0.0
4 0.0
5 0.0
6 0.0
7 0.0
# the integer values specified in $sdnc_c block indicate parameters

# in $sdnc block to be varied
$sdnc_c
7 1
1
1 1
2 2
3 3
4 4
5 5
6 6
7 7
# specification of vibrational transitions and their intensities

# in experimental rR spectra:
$rrs_exp
1 # number of rR spectra
1 1 # start of the block specifying the 1st rR spectrum
Ex 39809.0 # excitation energy for the first rR spectrum
NTr 7 # number of vibrational transitions for which intensities are
# provided
1
int 10.0 1.0
modes 1
quanta 1;
2
int 5.0 1.0

modes 2
quanta 1;
3
int 1.5 1.0
modes 3
quanta 1;
4
int 21.0 1.0
modes 4
quanta 1;
5
int 7.5 1.0
modes 5
quanta 1;
6
int 2.0 1.0
modes 6
quanta 1;
7
int 46.0 1.0
modes 7
quanta 1;
The input of rR intensities for an arbitrary number of excitation energies follows the keyword $rrs exp
block:
$rrs_exp
1 1
The first 1 in the last line denotes the number of the rR spectrum for which specification starts below. If
the second number is the same as the number of the spectrum, then it means that only relative intensities
for the first rR spectrum are meaningful in the fit. If several spectra are given in the input then the second
number may have a different value, e.g.:
$rrs_exp
1 2
...
This input is to be interpreted as indicating that 3 rR spectra are provided and the relative intensities for the
first spectrum are given on the same scale as the second one that will be accounted for in the fit. The value
of the excitation energies and the number of vibrational transitions specified are indispensable within the
blocks specifying intensities for each rR spectrum.
Following the number of vibrational transitions given by the keyword NTr one has to specify each vibrational
transition and its intensity. Thus, in the present case there are seven subblocks with the following structure:
k int I W
modes m1,m2,...mn
quanta q1,q2,...qn;
This means that the k-th transition has intensity I and weight W in the mean absolute difference function
that is used for the minimization (W is an optional parameter). The following 2 lines specify the vibrational
transitions by providing excitation numbers qi for modes mi so that the corresponding Raman shift is equal
P
to = , where i is vibrational frequency of the mode mi .
qi i
The parameters that are to be varied are specified within the constraint blocks $el states c and $sdnc c.
Both blocks have the same structure and number of parameters as $el states and $sdnc, respectively. A
parameter from the $el states block is supposed to be independently varied if its counterpart from the
$el states c block is equal to the number of the electronic state. Likewise, a parameter from the $sdnc
block is supposed to be independently varied if its counterpart from the $sdnc c block is equal to the number
of the normal mode. Model parameters that are set to 0 in the corresponding constraint blocks are not varied
in the fit. The values of the following parameters may be important for the quality of the fit:
CWAR in the %fit block specifies the weight of absorption relative to the weight of rR intensities in
the difference function to be minimized. If this parameter was not specified the fit would be almost
insensitive to the rR intensities in the input, since typically the number of experimental absorption
points is much larger than the number of rR transitions in the input. In most cases the value of CWAR
in the range 1.05.0 is a good choice since the error in the measured experimental intensity is expected
to be much smaller for absorption than for resonance Raman.
SDNCStep in the %fit block specifies the initial dimension of the simplex in the space of {m } and
should roughly correspond to the expected uncertainty of initial guess on {m } in the $sdnc block
compared to their actual values. You can notice in the present example that if this parameter is too
large (>2.0) or too small (<0.4) the quality of the fit may significantly deteriorate
Although the default initial dimensions of the simplex have reasonable values for different types of
parameters it may turn out to be helpful in some cases to modify the default values:
FREQGStep 10.0 # ground-state vibrational frequencies

FREQEStep 10.0 # excited-state vibrational frequencies
E0Step 300.0 # transition energies
SSStep 20.0 # Stokes shift
TMStep 0.5 # electronic transition dipole moment
GammaStep 50.0 # homogeneous linewidth
E0SDStep 50.0 # inhomogeneous linewidth
SDNCStep 1.0 # origin shift along dimensionless normal coordinate
The fit run of orca asa on the file example008.inp will converge upon approximately 700 function evaluations
(for MWADRelTol=1e-4). The results of the fit will be stored in file example008.001.inp which has the same
structure as the input file example008.inp. Thus, if the fit is not satisfactory and/or it is not fully converged
it can be refined in subsequent orca asa run upon which file example008.002.inp will be created, and so
on. Some model parameters in intermediate files can be be additionally modified and/or some constraints
can be lifted if so desired. The output file example008.001.inp will contain fitted displacement parameters
{m }stored in the $sdnc block:
$sdnc
7 1
1
1 0.675000
2 -0.194484
3 -0.217527
4 0.811573
5 0.529420
6 -0.149991
7 1.314915
In the present example, these parameters are actually in very close agreement with those published for the
hexatriene molecule!
The overall quality of the fit is determined by the parameter MWAD which upon convergence reaches the
value of 0.027. The fitted rR intensities are presented in the commented lines next to the experimental rR
intensities in file example008.001.inp:
$rrs_exp
1
1 1 3.495285e+001
Ex 39809.00
NT 7
1
Int 10.0 1.0 # simulated intensity: 1.000982e+001
modes 1
quanta 1;
2
Int 5.0 1.0 # simulated intensity: 8.976285e-001
modes 2
quanta 1;
3
modes 3
quanta 1;
4
modes 4
quanta 1;
5
modes 5
quanta 1;
6
Int 2.0 1.0 # simulated intensity: 6.014466e-001
modes 6
quanta 1;
7
modes 7
quanta 1;
The file hex-abs.fit.dat will contain the experimental and fitted absorption spectra in ASCII format which
can be plotted in order to visualize the quality of absorption fit (Figure 9.42).
Figure 9.42: Experimental (black) and fitted (red) absorption spectrum corresponding to 1-1 Ag
1-1 Bu transition in 1,3,5 trans-hexatriene.
NOTE
The more experimental rR intensities are included in the analysis the more reliable is the fit. In
principle it is possible to obtain fully consistent results even if only a limited number of vibrational
transitions is provided. However, in such a case it is desirable to include into analysis at least a single
Raman transition involving the mode for which is to be determined.
The quality of the fit can be improved if the IMDHOFA model is invoked and excited-state vibrational
frequencies are allowed to vary.
Due to the initial guess and dimension of the simplex, as well as some peculiarities of the sim-
plex algorithm for function minimization, you can still refine the fit by rerunning orca asa on file
example008.001.inp that may lead to an even lower value of the parameter MWAD = 0.021, and
therefore to better agreement of experimental and fitted spectra (even though the previous run has
been claimed to be converged).
In this respect it appears to be wise to perform the fit in 3 steps:
1. Fit the preresonance region below the 0-0 vibronic band with a single Lorentzian band, from
which the adiabatic transition energy E0 , and homogeneous linewidth are obtained. The range
for fit of the absorption spectrum can be specified by the AbsRange keyword in the %fit block.
2. Fix E0 and , and optimize {m } fitting the entire spectral range and rR intensities.
3. Lift constraints on E0 and , and reoptimize simultaneously all parameters.

9.25.3.3 Example: Single-Mode Fit of Absorption and Fluorescence Spectra for 1-1 Ag
1-1 B2u Transition in Tetracene
In this section we provide an example and discuss the most important aspects of joint fit of fluorescence and
absorption spectra. Figure 9.43 displays the experimental emission and absorption spectra corresponding to
1-1 Ag 1-1 B2u transition in tetracene.
Figure 9.43: Deconvoluted absorption (red) and fluorescence (blue) spectra of tetracene in cyclo-
hexane upon the assumption of a single vibronically active mode. The black solid lines
represent experimental spectra.
Both spectra show pronounced effective vibrational progressions that are dominated by 3 and 5 peaks,
respectively. As can be shown on the basis of quantum chemical calculations this progression has essentially
multimode character. However, the experimental spectra can be well fitted under the assumption of a single
vibronically active mode. The input has the following structure:
#
# example009.inp
#
# Parameters to be varied:
# 1) adiabatic minima transition energy
# 2) homogeneous and inhomogeneous linewidths
# 3) normal mode frequency and corresponding dimensionless displacement of the
# excited-state origin
#
%sim
Model IMDHO
EnInput E0 # we assume adiabatic minima separation energies
end
%fit
Fit true # global flag to turn on the fit
AbsFit true # flag to include absorption spectrum into the fit
FlFit true # flag to include fluorescence spectrum into the fit
WeightsAdjust true
AbsRange 19000.0, 28000.0 # spectral range for absorption

# which will be included into the fit
FlRange 17800.0, 22300.0 # spectral range for absorption

# which will be considered in the fit
AbsName "absexp.dat" # name of the file containing experimental

# absorption spectrum in a simple two-column
# ASCII format
FlName "flexp.dat" # name of the file with experimental fluorescence spectrum
ExpAbsScaleMode Rel # flags indicating that only relative shapes of the

ExpFlScaleMode Rel # absorption and fluorescence bands will be fitted.
CWAF 1.000 # important parameter to have a balanced relative quality of fit

# of fluorescence and absorption

# algorithm
MWADRelTol= 0.0001 # Relative Tolerance of the Mean Weighted Absolute

TMStep 0.5 # initial step for the transition dipole moments

E0SDStep 500.0 # initial step for the inhomogeneous linewidth (Sigma)
FREQGStep 100.00 # initial step for the vibrational frequencies
E0Step 1000.0 # initial step for the transition energies
SSStep 10.0 # initial step for the Stokes shift
GammaStep 100 # initial step for the homogeneous linewidth
SDNCStep 0.5 # initial step for the displacement parameter
end
$el_states
2
1 21100.00 100.00 100.00 1.0000 0.0000 0.0000
2 24000.00 100.00 1000.00 1.0000 0.0000 0.0000
$el_states_c
2
1 1 1 1 0 0 0
2 2 2 2 2 0 0
$abs_bool
2
1 1
2 1
$fl_bool
2
1 1
2 0
$ss
2
1 100.000000
2 0.000000
$ss_c
2
1 1
2 0
$vib_freq_gs
1
1 1500.0
$vib_freq_gs_c
1
1 1
$sdnc
1 2
1 2
1 2.0000000 0.000000
$sdnc_c
1 2
1 2
1 1 0
The parameter CWAF=1.0 in the %fit block specifies the weight of absorption relative to the weight of
fluorescence in the difference function to be minimized. If this parameter was not specified the quality of
the fit would be biased towards the spectrum with a larger number of experimental points. In some typical
situations where the error in the measured experimental intensity is expected to be smaller for absorption
than for emission it is desirable to choose the value of CWAF to be more than 1.0.
In order to account for a broad featureless background signal in the absorption spectrum above 24000 cm1 ,
the second band was included into the analysis and approximated with a Voigt lineshape which means also
that the corresponding frequency in the $vib freq gs block and displacement parameter in the $sdnc block
are fixed to zero in the fit. Thus, the $el states block contains an initial guess on the transition energies,
transition electric dipole moments and linewidth parameters for 2 states:
$el_states
2
1 21100.00 100.00 100.00 1.0000 0.0000 0.0000
2 24000.00 100.00 1000.00 1.0000 0.0000 0.0000
The initial value of the adiabatic minima separation energy for the first state was approximated by the
energy corresponding to the first vibronic peak in the absorption spectrum (21100 cm1 ). The transition
energies and linewidth parameters are varied independently as indicated in the $el states c block. Since
we allow to fit only bandshapes, but not the overall intensities of the spectra, only relative absolute values
of the transition electric dipole moments of two bands are important. Therefore it is reasonable to fix all
components of the transition moment for the first state and vary only Mx component for the second one:
$el_states_c
2
1 1 1 1 0 0 0
2 2 2 2 2 0 0
Since we assume the absorption by both states and emission only from the first one, it is necessary to include
Boolean arrays $abs bool and $fl bool which specify states which will be included in the treatment of the
absorption and fluorescence spectra, respectively:
$abs_bool
2
1 1 # 1 indicates that the corresponding state will be included in the calculation of
2 1 # absorption
$fl_bool
2
1 1
2 0 # 0 indicates that the corresponding state will be excluded from the calculation
# of emission spectrum
We need also to vary the value of vibrational frequency of the mode which determines separation of vibrational
peaks in the spectra. This is done via the constraint block $vib freq gs c:
$vib_freq_gs_c
1
1 1
Note that it is meaningless to include into the treatment the Stokes shift for the second state which give rise
to the background signal in the absorption since the corresponding emission is not present. Therefore for
the second state is fixed to zero as indicated in the $ss block and its constraint counterpart $ss c:
$ss
2
1 100.000000 # initialization of the Stokes shift for the 1st electronic state
2 0.000000
$ss_c
2
1 1 # the Stokes shift for the 1st electronic state will be varied in the fit
2 0 # the Stokes shift for the 2nd electronic state will be fixed in the fit
The fit run of orca asa on file example009.inp will converge upon approximately 700 function evaluations (for
MWADRelTol=1e-4). The file example009.001.inp will contain the fitted effective values of the vibrational
frequency and dimensionless displacement: = 1404 cm1 , = 1.35. One can notice that the fit is rather
poor in the low- and high-energy edges of the absorption and fluorescence spectra, respectively (Figure 9.43).
The source of this discrepancy is the single-mode approximation which was employed here. The quality of the
fit can be significantly improved assuming several modes with non-zero displacement parameters. Note that
in such a case the proper guess on the number of active modes and corresponding dimensionless displacements
can be deduced from quantum chemical calculations.
9.25.4 Quantum-Chemically Assisted Simulations and Fits of Optical Bandshapes

and Resonance Raman Intensities
In this section we finally connect the spectra simulation algorithms to actual quantum chemical calculations
and outline a detailed approach for the analysis of absorption, fluorescence and resonance Raman spectra
within the IMDHO model. Our procedure becomes highly efficient and nearly automatic if analytical
excited state derivatives with respect to nuclear displacements are available. However, this availability is not
mandatory and hence, spectral predictions may as well be achieved by means of normal mode scan calculations
for high-level electronic structure methods for which analytic gradients have not been implemented.
9.25.4.1 Example: Quantum-Chemically Assisted Analysis and Fit of the Absorption and
Resonance Raman Spectra for 1-1 Ag 1-1 Bu Transition in trans-1,3,5-Hexatriene
The following input file for an ORCA run invokes the calculation of the excited-state origin displacements
along all normal modes by means of energy and excited state gradient calculations at the ground-state
equilibrium geometry. The method is valid for the IMDHO model for which the excited-state energy gradient
along a given normal mode and corresponding origin shift are related in a very simple way.
#
# example010.inp
#
# TDDFT BHLYP Normal Mode Gradient Calculation
#
# The keyword NMGrad invokes the normal mode gradient calculation
#
! RKS BHandHLYP TightSCF SV(P) NMGrad
%cis NRoots 1
triplets false
end
%rr
# the nuclear Hessian must have been calculated before - for example by a
# DFT calculation.
HessName= "hexatriene.hess"
states 1 # Perform energy-gradient calculations for the 1st

# excited state.
Tdnc 0.005 # Threshold for dimensionless displacements to be
# included in the input file for spectral simulations
# generated at the end of the program run.
# By default Tdnc= 0.005
ASAInput true # Generate the input file for spectra simulations

end
* xyz 0 1
C -0.003374 0.678229 0.00000
H -0.969173 1.203538 0.00000
C 1.190547 1.505313 0.00000
H 2.151896 0.972469 0.00000
C 1.189404 2.852603 0.00000
H 0.251463 3.423183 0.00000
H 2.122793 3.426578 0.00000
C 0.003374 -0.678229 0.00000
H 0.969172 -1.203538 0.00000
C -1.190547 -1.505313 0.00000
H -2.151897 -0.972469 0.00000
C -1.189404 -2.852603 0.00000
H -0.251463 -3.423183 0.00000
H -2.122793 -3.426578 0.00000
*
In the ORCA run the TDDFT excited state gradient calculations are performed on top of a TDDFT
calculation. Note, that the numbers of the excited-states which have to be included into analysis and input
file for spectral simulations must be specified after the States keyword in the %rr block. They should also
be consistent with the required number of roots in the %tddft block. The 1-1 Bu excited state appears to
be the first root in the TDDFT calculation. Therefore, NRoots=1 in the %tddft block, and States=1 in
the %rr block. One should also provide the name of the file containing the nuclear Hessian matrix via the
HessName keyword in the %rr block. Here we used the .hess file obtained in a frequency calculation at the
BHLYP/SV(P) level of theory.
After the ORCA calculation you will find in your directory a file called example010.asa.inp that is
appropriate to be used together with the orca asa program as defined in the preceding sections.
#
# example010.asa.inp
#
# ASA input
#
%sim
model IMDHO
method Heller
AbsRange 5000.0, 100000.0
NAbsPoints 0
FlRange 5000.0, 100000.0

NFlPoints 0
RRPRange 5000.0, 100000.0

NRRPPoints 0
RRSRange 0.0, 4000.0

NRRSPoints 4000
RRS_FWHM 10.0
AbsScaleMode Ext
FlScaleMode Rel
RamanOrder 0
EnInput E0
CAR 0.800
end
%fit
Fit false
AbsFit false
FlFit false
RRPFit fsalse
RRSFit false
method Simplex
WeightsAdjust true
AbsRange 0.0, 10000000.0

FlRange 0.0, 10000000.0
RRPRange 0.0, 10000000.0
RRSRange 0.0, 10000000.0
AbsName ""
FlName ""
ExpFlScaleMode Rel
ExpAbsScaleMode Rel
CWAR -1.000
CWAF -1.000
NMaxFunc 100
MWADRelTol= 1.000000e-004
SFRRPSimStep= 1.000000e+002
SFRRSSimStep 1.000000e+002
FREQGStep 1.000000e+001
FREQEStep 1.000000e+001
E0Step 3.000000e+002
SSStep 2.000000e+001
TMStep 5.000000e-001
GammaStep 5.000000e+001
E0SDStep 5.000000e+001
SDNCStep 4.000000e-001
end
$el_states
1
1 42671.71 100.00 0.00 1.0725 3.3770 -0.0000
$vib_freq_gs
12
1 359.709864
2 456.925612
3 974.521651
4 1259.779018
5 1356.134238
6 1370.721341
7 1476.878592
8 1724.259894
9 1804.572974
10 3236.588264
11 3244.034359
12 3323.831066
$sdnc
12 1
1
1 -0.594359
2 0.369227
3 -0.132430
4 -0.727616
5 0.406841
6 -0.105324
7 0.177617
8 -0.090105
9 -1.412258
10 0.048788
11 0.021438
12 0.008887
This input file can be used to construct theoretical absorption and rR spectra. In order to compare
experimental and theoretical rR spectra, it is necessary to use in both cases excitation energies
that are approximately in resonance with the same vibrational transitions in the absorption
spectrum. Therefore, in the case of the absorption spectrum with resolved or partially re-
solved vibrational structure it is necessary to modify the transition energies in the %el states
such that they coincide with the experimentally observed 0-0 vibrational peaks. It is also de-
sirable to roughly adjust homogeneous and, possibly, inhomogeneous linewidth parameters such that the
experimental and calculated absorption spectra show similar slopes in the preresonance region (below the
0-0 transition). Then the assignment of experimental rR spectra can be done on the basis of comparison
with the theoretical rR spectra calculated for the corresponding experimental excitation energies. For the
sake of consistency and simplicity it is better to use those excitation energies which fall into the preresonace
region and/or are in resonance with the 0-0 transition. In the case of diffuse absorption spectra (i.e.
those not showing resolved vibrational structure) it is also necessary to adjust the theoretical
transition energies and linewidth parameters such that experimental and calculated positions
of absorption maxima roughly coincide, and corresponding slopes below the maxima have a
similar behavior. According to above mentioned considerations one needs to modify the %el states block
in the file example010.asa.inp:
$el_states
1
1 39808.0 150.00 0.00 1.0725 3.3770 -0.0000
The calculated absorption spectrum obtained by providing AbsScaleMode= Rel, AbsRange= 39000, 49000
and NAbsPoints= 2000 is shown in Figure 9.44. Upon comparison with the experimental spectrum one can
notice that the BHLYP functional gives relatively small discrepancies with somewhat lower intensity in the
low-frequency edge and larger intensity on the high-energy side of the spectrum. Besides, there is a noticeable
mismatch in the separation between individual vibronic peaks which is due to overestimation of vibrational
frequencies by the BHLYP functional (typically by 10%).
You can arbitrarily vary various normal coordinate displacements in %sdnc block within 1030% of their values
in order to observe modifications of the calculated spectrum. This will tell you how these parameters influence
the spectrum and probably it will be possible to obtain better initial guesses for the fit. In the present
example you will find that reduction of the absolute value of the displacement parameter corresponding
to the ninth mode by 10%, and reduction of vibrational frequencies by 10% can noticeably improve
the spectral envelope. Such a quick analysis suggests that experimentally observed peaks in the absorption
spectrum represent different vibrational transitions corresponding to a single electronically excited state
rather than to different electronic excitations. This conclusion will be confirmed upon establishing the fact
that the absorption and rR spectra can be successfully fitted based on the assumption of a single electronic
transition.
In order to calculate the rR spectrum for experimental excitation energies you need to specify its value
through RRSE keyword in %sim block as well as possibly to modify the parameters related to the spectral
range and linewidth of rR bands which are suitable for comparison with the experimental rR spectrum:
# excitation energies (cm**-1) for which rR spectra will be calculated:

RRSE 39808
# full width half maximum of Raman bands in rR spectra (cm**-1):

RRS_FWHM 20
RSRange 0, 4000 # spectral range for simulation of rR spectra (cm**-1)

NRRSPoints 4000 # number of points to simulate rR spectra (cm**-1)
# resonance Raman intensities will be calculated

# for all vibrational states with excitation number
# up to RamanOrder:
RamanOrder 3
Figure 9.44: Experimental and calculated at the BHLYP/SV(P) and B3LYP/SV(P) levels of theory
absorption (left panel) and rR spectra (right panel) corresponding to 1-1 Ag 1-1 Bu
transition in trans-1,3,5-hexatriene.
The calculated rR spectrum is shown in Figure 9.44. In the input we have invoked the calculation of rR
intensities for the transitions with up to 3 vibrational quanta in the final vibrational state (RamanOrder =
3). Make sure that the rR intensity pattern in the given spectral range does not change noticeably upon
further increase of this parameter. Typically, the larger are the normal coordinate displacements the greater
order of Raman scattering is required in the calculation to account for all the most intense transitions in the
rR spectrum. The inclusion of vibrational transitions beyond the fundamentals is a particular feature of the
orca asa program.
Comparison of the calculated and experimental rR spectra (Figure 9.44) mainly shows discrepancies in
the values of the Raman shifts that are mainly related to the low accuracy of the vibrational frequencies
obtained at the BHLYP level (typically overestimated by 10%). However, the intensity patterns of the
calculated and experimental rR spectra show very nice agreement with experiment that is already sufficient
to assign the experimental peaks to individual vibrational transitions. This can be done upon examination of
file example010.asa.o3.rrs.39808.stk which provides intensity, Raman shift, and specification for each
vibrational transition. It is actually one of the most consistent procedures that enables one to identify
different fundamental, overtone and combination bands in the experimentally observed rR spectrum. Such an
assignment is a necessary prerequisite for the fit. The current example is relatively straightforward since the
spectral region 11700 cm1 is actually dominated by fundamental bands while the most intense overtone and
combination transitions occur at higher frequencies. However, in many cases even the low-frequency spectral
range is characterized by significant contributions from overtone and combination bands that sometimes are
even more intense than fundamental transitions! Thus, quantum chemical calculations can greatly facilitate
the assignment of experimental rR bands.
After having performed the assignment it is advisable to discard those modes from the analysis that are
not involved in any of the experimentally observed fundamental, overtone, or combination rR bands with
noticeable intensities. In the present example these are the modes 6, 8, 1012 from the input file given
above. For these modes it is implied that the fitted displacement parameters are zero. You will find that
the calculated displacement values are rather small indeed. Also it is advisable to change the ground-state
vibrational frequencies in the $vib freq gs block to their experimental values.
Below is the modified input file for the fit run:
#
# example010-01.asa.inp
#
# ASA input
#
%sim
model IMDHO
method Heller
end
%fit
Fit true
AbsFit true
RRSFit true
AbsExpName "hex-abs.dat"
ExpAbsScaleMode Rel
CWAR 5.0
NMaxFunc 1000
SDNCStep 0.5
end
$el_states
1
1 39808.0 150.00 0.00 -0.8533 -3.3690 -0.0000
$el_states_c
1
1 1 1 0 0 0 0
$vib_freq_gs
7
1 354.0
2 444.0
3 934.0
4 1192.0
5 1290.0
6 1403.0
7 1635.0
$sdnc
7 1
1
1 -0.594359
2 0.369227
3 -0.132430
4 -0.727616
5 0.406841
6 0.177617
7 -1.412258
$sdnc_c
7 1
1
1 1
2 2
3 3
4 4
5 5
6 6
7 7
$rrs_exp
1
1 1
Ex 39809.0
NTr 11
1
int 10.0 1.0
modes 1
quanta 1;
2
int 5.0 1.0
modes 2
quanta 1;
3
int 1.5 1.0
modes 3
quanta 1;
4
int 21.0 1.0
modes 4
quanta 1;
5
int 7.5 1.0
modes 5
quanta 1;
6
int 2.0 1.0
modes 6
quanta 1;
7
int 46.0 1.0
modes 7
quanta 1;
8
int 6.8 1.0
modes 1, 7
quanta 1, 1;
9
int 4.0 1.0
modes 2, 7
quanta 1, 1;
10
int 2.0 1.0
modes 3, 7
quanta 1, 1;
11
int 17.0 1.0
modes 7
quanta 2;
In addition to the experimental intensities of fundamental bands the input file also contains the information
about some overtone and combination transitions. Note that it is not really necessary to include all of them
them into the fit, in particular if some of the rR bands are strongly overlapping with each other.
Fitted normal coordinate displacements of the excited-state origin show nice agreement with the published
values:
$sdnc
7 1
1
1 -0.638244
2 0.455355
3 -0.229126
4 -0.854357
5 0.501219
6 0.197679
7 -1.292997
NOTE
It is not really important to employ the BHLYP/SV(P) method in the frequency calculations in order to
obtain the .hess file (this was merely done to be consistent with the TDDFT/BHLYP/SV(P) method
for the excited-state model parameters calculation). The frequency calculations can for example be
carried out at the BP86/TZVP or RI-SCS-MP2/TZVP level of theory. This will provide displacements
pattern very similar to that of the BHLYP/SV(P) method, but much more accurate vibrational
frequencies which will further facilitate the assignment of rR spectra (Figure 9.44). However, such a
procedure can be inconsistent if the two methods give noticeably different normal mode compositions
and/or vibrational frequencies. From our experience it can lead to significant overestimation of the
excited-state displacements for some low-frequency modes.
It is known that predicted dimensionless normal coordinate displacements critically depend on the
fraction of the exact Hartree-Fock exchange (EEX) included in hybrid functionals. In general no
universal amount of EEX exists that provides a uniformly good description for all systems and states.
Typically, for a given molecule either the BHLYP/TZVP (50% of EEX) or B3LYP/TZVP (20% of EEX)
methods yields simulated spectra that compare very well with those from experiment if vibrational
frequencies are appropriately scaled.
9.25.4.2 Important Notes about Proper Comparison of Experimental and Quantum

Chemically Calculated Resonance Raman Spectra
In order to compare experimental and theoretical rR spectra, it is necessary to use in both cases
excitation energies that are approximately in resonance with the same vibrational transitions
in the absorption spectrum. Therefore, in the case of diffuse absorption spectra (i.e. those not showing
resolved vibrational structure) one needs to adjust the transition energies and linewidth paramters in the
%el states block such that the envelopes of the experimental and theoretical spectra rouhgly coincide, and
then to employ experimental values of excitation energies to construct theoretical rR spectra. Typically
in the case of diffuse absorption spectra rR profiles are rather smooth. Therefore, even though excitation
energies are not in resonance with the same vibrational transition in the absorption spectrum, the rR spectra
are not expected to vary significantly in the case of such mismatch.
In the case of the absorption spectrum with resolved or partially resolved vibrational structure it is necessary
to modify the transition energies in the %el states block such that the calculated and experimentally
observed 0-0 vibrational peaks coincide, and modify linewidth parameters so that the low-energy slopes in
the calculated and experimental spectra have a similar behavior.
Consider a single-mode model system for which experimental and calculated absorption spectra are shown
in Figure 9.45.
Comaprison of the calculated and experimental spectra shows that some adjustment of the linewidth
parameters is neceassy before construction of theoretical rR spectra. One can directly compare calculated and
experimental rR spectra upon the excitation at 16200 cm1 which is in resonance with the 0-0 vibronic band.
However, it is not consistent to use experimental values of the excitation energy in the calculation of rR
spectrum which is in resonance with one of the other vibronic bands since the separation between vibrational
peaks in the experimental and calculated spectra is different whereby positions of the peaks in both spectra
do not coincide. Instead one should use the excitaition energy which corresponds to the same
vibronic peak in the calculated absorption spectrum as in the experimental one. Alternatively,
one can adjust theoretical value of vibrational frequency such that positions of corresponding vibronic peaks
in the spectra coincide, and then use experimental values of excitation energies for the calculation of rR
spectra.
9.25.4.3 Example: Normal Mode Scan Calculations of Model Parameters for 1-1 Ag 1-1 Bu
Transition in trans-1,3,5-Hexatriene
If excited state gradients are not available (which is the case for many of the electronic structure methods
supported by ORCA), you have to resort to a more laborious procedure single point calculations at
Figure 9.45: Experimental and theoretical absorption spectra for a single-mode model system. The
calculated spectrum is adjusted such that the position of 0-0 peak coincide with the
experimental one.
geometries that are displaced along the various normal modes of the system. This roughly corresponds to
taking numerical derivatives however, once this extra effort is invested more information can be obtained
from the calculation than what would be possible from an analytic derivative calculation.
The present example illustrates the application of normal mode scan calculations for the evaluation of excited
state harmonic parameters that are necessary to simulate optical spectra within the IMDHO model. This
method can be applied with any method like CIS, CASSCF, MRCI or TD-DFT.
The reference wavefunctions for the multireference calculations reported below are of the state-averaged
CASSCF (SA-CASSCF) type. The complete active space CAS(6,6) includes all 6 valence shell -orbitals.
The average is taken over the first four states which was found necessary in order to include the ground state
and the strongly allowed 1-1 Bu state.
#
# example011.inp
#
# CASSCF normal mode scan calculations
#
# first do single point RHF calculation

! RHF TZVP TightSCF
* xyz 0 1
C -0.002759 0.680006 0.000000
H -0.966741 1.204366 0.000000
C 1.187413 1.500920 0.000000

H 2.146702 0.969304 0.000000
C 1.187413 2.850514 0.000000
H 0.254386 3.420500 0.000000
H 2.116263 3.422544 0.000000
C 0.002759 -0.680006 0.000000
H 0.966741 -1.204366 0.000000
C -1.187413 -1.500920 0.000000
H -2.146702 -0.969304 0.000000
C -1.187413 -2.850514 0.000000
H -0.254386 -3.420500 0.000000
H -2.116263 -3.422544 0.000000
*
# perform SA-CASSCF calculation upon appropriate rotation of MOs

$new_job
! RHF TZVP TightSCF
%scf
rotate {23,27,90} end
end
%casscf
nel 6
norb 6
mult 1
nroots 4
end
* xyz 0 1
C -0.002759 0.680006 0.000000
H -0.966741 1.204366 0.000000
C 1.187413 1.500920 0.000000
H 2.146702 0.969304 0.000000
C 1.187413 2.850514 0.000000
H 0.254386 3.420500 0.000000
H 2.116263 3.422544 0.000000
C 0.002759 -0.680006 0.000000
H 0.966741 -1.204366 0.000000
C -1.187413 -1.500920 0.000000
H -2.146702 -0.969304 0.000000
C -1.187413 -2.850514 0.000000
H -0.254386 -3.420500 0.000000
H -2.116263 -3.422544 0.000000
*
# do normal mode scan calculations

# to map CASSCF ground and excited-state PESs
$new_job
! RHF TZVP TightSCF NMScan
%casscf
nel 6
norb 6
mult 1
nroots 4
end
%rr
HessName "hexatriene_bp86.hess"
NMList 10,11,18,24,26,28,29,31,32
NSteps 6
FreqAlter true
EnStep 0.0001
State 3
end
* xyz 0 1
C -0.002759 0.680006 0.000000
H -0.966741 1.204366 0.000000
C 1.187413 1.500920 0.000000
H 2.146702 0.969304 0.000000
C 1.187413 2.850514 0.000000
H 0.254386 3.420500 0.000000
H 2.116263 3.422544 0.000000
C 0.002759 -0.680006 0.000000
H 0.966741 -1.204366 0.000000
C -1.187413 -1.500920 0.000000
H -2.146702 -0.969304 0.000000
C -1.187413 -2.850514 0.000000
H -0.254386 -3.420500 0.000000
H -2.116263 -3.422544 0.000000
*
The file containing the hessian matrix ("hexatriene bp86.hess") was obtained from the BP86/TZVP
frequency calculations. The keyword NMList provides the list of the normal modes to be scanned. These
should be only the totally symmetric vibrations, since only they can be significant for absorption and resonance
Raman spectra within the constraints of the IMDHO model. The FreqAlter flag indicates whether frequency
alterations are assumed in the post-scan potential surface fit. The Parameter EnStep is used to select the
appropriate step during the scan calculations. The value is chosen such that the average energy change (in
Eh) in both directions is not less than this parameter.
9.26 Ab initio Molecular Dynamics Simulations 665
9.26 Ab initio Molecular Dynamics Simulations
Recently, it was decided to include an ab initio molecular dynamics (AIMD) module in ORCA.19 As a plethora
of different electron structure methods with analytical gradients is already implemented, all these methods
are now available also for MD simulations, offering a wide range of accuracy/performance tradeoffs.
This is the first version of ORCA which contains the MD module, and the code is not yet mature. Only a
small subset of features one would typically expect in such a module is currently implemented. In future
releases, many new features and methods will (hopefully) be added to this part of the program. However, we
will do our best to keep a strict backward compatibility, such that the sample inputs from this section will
remain valid in all future releases.
9.26.1 Input Format
The molecular dynamics module is activated by specifying MD in the simple input line. The actual MD
input which describes the simulation follows in the %md section at some later position in the input file. The
contents of this section will subsequently be referred to as MD input.
! MD PBE def2-SVP
%md
timestep 0.5 fs
initvel 350 k
thermostat berendsen 350 k timecon 10 fs
dump position stride 1 file "trajectory.xyz"
run 200
end
* xyz 0 1
O -4.54021 0.78439 0.09307
H -3.64059 0.38224 -0.01432
H -4.63463 1.39665 -0.67880
*
Please note that the MD input is not processed by ORCAs main parser, but by a dedicated parser in the
MD module. Therefore, the MD input is not required to obey the general ORCA syntax rules. The syntax
will be described in the following.
In contrast to general ORCA input, the MD input is not based on keywords, but on commands, which are
executed consecutively on a line-by-line basis starting at the top (like, e. g., in a shell script). This means
that identical commands with different arguments may be given, coming into effect when the interpreter
reaches the corresponding line. This enables to perform multiple simulations (e. g., pre-equilibration and
production run) within a single input file:
19
Strictly speaking, these simulations are BornOppenheimer molecular dynamics simulations (BOMD), because they
approximately solve the time-independent Schr odinger equation to compute gradients and then move the atoms
according to these gradients.
%md
timestep 1.0 fs
run 200
timestep 2.0 fs
run 500
end
At some point in the future, it is planned to add variable definitions, loops, and conditional branching to the
MD input.20 This will enable even larger flexibility (e. g., to run a simulation until a certain quantity has
converged).
Like in standard ORCA input, comments in the MD input are initiated by a # sign and span to the end
of the current line. Commands can be started both at the beginning of a line and after a command. The only
place where a # is not treated as start of a comment is inside of a string literal (e. g., in file names).
%md
# Comment
timestep 0.5 fs # Comment
dump position file "trajectory#1.xyz"
end
Some more MD input syntax rules:
The MD input is generally not case-sensitive. The only exception are file names on platforms with
case-sensitive file systems (Linux).
Empty lines are allowed.
Commands and options are separated by space or tabulator characters. Any combination of these
characters may be used as separator.
Both DOS and UNIX line break style is acceptable.
Multiple commands in the same line are not allowed.
A command with a long argument list may not span over multiple lines (maybe this will be relaxed in
the future).
20
Technically speaking, ORCA will then be a turing-complete script interpreter, such that any computational problem
can be solved with ORCA :-)
Commands
As already noted above, the central item of the MD input is a command. Each input line contains (at most)
one command, and these commands are executed in the given order. A command typically takes one or more
arguments, which are given behind the command name, separated by whitespaces or tabulator characters.
The order of the arguments for a command is fixed (see command list in section 9.26.2). Commands may have
optional arguments, which are always specified at the end of the argument list, after the last non-optional
argument. If there exist multiple optional arguments for a command, not all of them need to be specified;
however, they need to be specified in the correct order and without gaps:
%md
command arg1 arg2 arg3 # fine
command arg1 arg2 arg3 optarg1 # fine
command arg1 arg2 arg3 optarg1 optarg2 # fine
command arg1 arg2 arg3 optarg2 # will not work
end
Apart from arguments and optional arguments, commands can also have modifiers. These can be considered
as sub-commands, which modify a given command, and may possess their own argument lists. Modifiers
generally follow after all non-optional and optional arguments, and they may not possess optional arguments
on their own. If a command has multiple modifiers, the order in which they are given is not important.
In the following input example, mod1 and mod2 are modifiers of command. mod1 takes one argument,
mod2 does not take arguments:
%md
command arg1 # fine
command arg1 optarg1 # fine
command arg1 mod1 modarg1 mod2 # fine
command arg1 mod2 # fine
command arg1 mod2 mod1 modarg1 # fine
command arg1 optarg1 mod1 modarg1 mod2 # fine
end
To make this abstract definition a little more illustrative, please consider again one line from the input sample
at the beginning of this section:
%md
dump position stride 1 file "trajectory.xyz"
end
Here, dump is the command, which takes one non-optional argument to specify which quantity shall be
dumped - in this case, position. The dump command has two modifiers, namely stride and file.
The former takes one integer argument, the latter a string argument. Swapping the two modifiers (together
with their respective arguments, of course) would not change the behavior.
Physical Units
In many cases, it is required to specify quantities which bear a physical unit in an input file (e. g., temperature,
time step lengths, . . . ). For many quantities, there are different units in widespread use, which always leads
to some confusion (just consider the kcal vs kJ case). ORCA handles this problem by defining default
units for each quantity and requiring that all quantities are given in their default unit. ORCAs default
units are the atomic units, which are heavily used in the quantum chemistry community, but not so much in
the molecular dynamics community. As an ab initio molecular dynamics module exists in the small overlap
region of both communities, some unit conflicts might arise. To prevent those from the beginning, it is
allowed to specify units of personal choice within ORCAs MD input.
Luckily, this is as simple and convenient as it sounds. The parser of the MD module checks if a unit is given
after a numeric constant, and automatically converts the constant to the internal default unit. If no explicit
unit is given, the default unit is assumed. Please note that the default units within the MD module are not
necessarily atomic units (see table below).
%md
timestep 1.0 fs
timestep 41.3 au # identical
timestep 1.0 # identical, as default time unit in MD module is fs
end
In the following, all units which are currently known to the MD modules parser are listed, sorted by physical
quantities. The default unit for each quantity is printed in bold letters. Additive constant and factor are
applied to convert a unit into the default unit. The additive constant is applied before the factor. A - sign
means that the constant/factor is not applied. More units will be probably added in the future.
Unit Symbol Additive constant Factor

Length Units
Angstrom - -
A - -
Bohr 0.5291
Time Units
fs - -
ps - 1000
au - 0.02419
Temperature Units
Kelvin - -
K - -
Celsius 273.15 -
C 273.15 -
9.26.2 Command List
In the following, a list of all commands currently known to the MD module will be given. The description
of each command starts with a small box which contains the commands name and a table of arguments
and modifiers. The last-but-one column in the table specifies the type of each argument. Possible types
are integer, real, string, and keyword. In the latter case, the last column contains a list of allowed
keyword values in { braces }. If the type is real and is a physical quantity with unit, the quantity is given
in the last column in [square brackets]. This table is followed by a text description of the command.
timestep
Mandatory Arguments: dt real [time]
Optional Arguments: -
Modifiers: -
Sets the simulation time step used to integrate the equations of motion for all following runs to dt. If
your system contains hydrogen atoms, a time step not above 0.5 fs is recommended. If only heavier
atoms are present, a larger time step may be chosen.
If this command is not invoked before a run call, a default time step of 0.5 fs will be set before starting
the run.
initvel
Mandatory Arguments: temp real [temperature]
Modifiers: -
Initializes the velocities of the atoms by random numbers based on a MaxwellBoltzmann distribution,
such that the initial temperature matches temp (see also section 9.26.3.2). Please note that this overwrites
all velocities, so do not call this command when your system is already equilibrated (e. g., to change
temperature use a thermostat instead).
The total linear momentum of the initial configuration is automatically removed, such that the system
will not start to drift away when the simulation begins. This only concerns the initial configuration.
Total linear momentum might build up during the simulation due to numeric effects.
If this command is not invoked before a run call, the atom velocities will be initialized to zero before
starting the run.
thermostat
Mandatory Arguments: type keyword { berendsen, none }
Optional Arguments: temperature real [temperature]
Modifiers: timecon dt real [time]
massive -
Changes the atom thermostat settings for subsequent simulation runs. type sets the thermostat type.
Currently, only the Berendsen thermostat is implemented. Use none as type to disable the thermostat.
The temperature argument sets the target temperature to which the system is thermostated. If this
argument is omitted, the temperature from the last call to the initvel command is used (if no such call
was invoked before, the simulation is aborted).
The timecon modifier sets the coupling strength of the thermostat (large time constants correspond to
weak coupling). The default value is 10 fs, which is a relatively strong coupling. Values in the range of
10 . . . 100 fs are reasonable (see also section 9.26.3.3).
The massive modifier activates massive thermostating, which means that each degree of freedom is
assigned to an independent thermostat. This is useful for pre-equilibration runs (helps to reach energy
equipartition) and should not be used during production runs, as it heavily distorts the dynamics.
Please note that a Berendsen thermostat will show no effect if the systems temperature is 0 K, as it
works by multiplying the velocities with a factor.
dump
Mandatory Arguments: quantity keyword { position, velocity, force }
Modifiers: format fmt keyword { xyz }
stride n integer -
file fname string -
none -
Specifies how to write the output trajectory of the simulation. The quantity argument can be one of the
keywords position, velocity, and force.
The stride modifier specifies to write only every n-th time step to the output file (default is n = 1).
The format modifier sets the format of the output file. Currently, only the xyz format is implemented,
which is also the default format.
The file modifier gives the output file name.
Use the none modifier to disable writing this quantity to an output file.
The default is to write a position trajectory with stride 1 and format xyz to a file named proj-
postrj.xyz, where proj is the base name of the ORCA project. If you want to create no output
trajectory at all, use dump position none.
printlevel
Mandatory Arguments: value keyword { low, medium, high, debug }
Modifiers: -
Controls the amount of information which is printed to the screen during the simulation. debug should
be used only in rare cases, because it might slow the simulation down heavily.
The default value is medium.
scflog
Mandatory Arguments: value keyword { discard, last, append, each }
Modifiers: -
Controls how/if the detailed output from the electron structure calculation (i. e., integrals, scf, gradient,
. . . ) will be written to log files. discard completely discards the output. last only keeps the last output
for each program call (useful to read error message if simulation aborts). append redirects all the output
into one single log file, appending each step at the end of the file. each writes the output for each step
and each program to different log files, which have the step number in their file names.
To print the detailed output of the electron structure calculation to the screen instead, see the printlevel
command.
The default value is append.
run
Mandatory Arguments: n integer -
Modifiers: -
Performs a simulation run over n time steps with the current settings, applying the Velocity Verlet
algorithm to solve the equations of motion (see section 9.26.3.1). You might want to call commands like
timestep, initvel, thermostat, and dump before.
If no call to initvel occurred before this command, the atom velocities are initialized to zero.
If no call to timestep occurred before this command, a default time step of 0.5 fs is set.
9.26.3 Scientific Background
In this section, some of the methods and algorithms used within ORCAs MD module are described in some
more depth, with a focus on the scientific background.
9.26.3.1 Time Integration and Equations of Motion
The central concept of molecular dynamics simulations is to solve Newtons equations of motion (at least as
long as the atom cores are treated classically). These read

Fi ~x (t)
x
i (t) = , i = 1 . . . N, (9.309)
mi
where xi (t) denotes the position of the i-th degree of freedom at time t, m the corresponding mass, and Fi
the force acting upon this degree of freedom. As the force may depend on all positions, this is a coupled
system of N ordinary differential equations (ODEs). In the general case, it is not possible to obtain an
analytical solution of this system, and therefore numerical solution methods are applied. These are almost
always based on discretizing the time variable and approximately solving the system by taking finite time
steps.
Of all different methods to numerically solve coupled systems of ODEs, the symplectic integration schemes
for Hamiltonian systems attained special attention in the field of molecular dynamics. They possess a very
good conservation of energy. In contrast to many other methods, they show a reasonable behavior when
investigating the long-term evolution of chaotic Hamiltonian systems (like, e. g., MD simulations). Three
popular such symplectic integration schemes are the Leapfrog algorithm, the Verlet method, and the Velocity
Verlet integrator. Despite their different names, they are very similar. It can be easily seen that the Verlet
and Velocity Verlet methods are algebraically equivalent (by eliminating the velocities from the Velocity
Verlet algorithm), and it can be shown that, eventually, all three methods are identical.21 All three methods
are explicit integration methods with a global error of order 2, and therefore one order better than the
semi-implicit Euler method, which is also a symplectic integration scheme. As the Velocity Verlet algorithm
is the only of these three methods which yields velocities and positions at the same point in time, many
popular molecular dynamics packages (CP2k, CPMD, LAMMPS) use this scheme. For the same reasons, the
ORCA MD module uses the Velocity Verlet algorithm as time integration method.
The general equations of the Velocity Verlet scheme read
1
~x (t + t) = ~x (t) + ~v (t) t + ~a (t) t2 , (9.310)
2
~a (t) + ~a (t + t)
~v (t + t) = ~v (r) + t. (9.311)
2
By inserting
F~i (t)
~ai (t) = , i = 1 . . . N, (9.312)
mi
one arrives at the two-step method
F~i (t) 2
~xi (t + t) = ~xi (t) + ~vi (t) t + t , i = 1 . . . N, (9.313)
2mi
F~i (t) + F~i (t + t)
~vi (t + t) = ~vi (r) + t, i = 1 . . . N, (9.314)
2mi
21
Hairer, Lubich, Wanner, Geometric Numerical Integration, Springer 2006.
which is implemented in ORCAs MD module.
9.26.3.2 Velocity Initialization
In the beginning of a MD simulation, it is often the case that only the initial positions of the atoms are
known, but not the velocities. As MD simulations are performed at some finite temperature, it is a good
idea to initialize the velocities in a way such that the desired simulation temperature is already present in
the beginning. In statistical mechanics, it is often assumed that the velocity distribution of atoms is given
by a MaxwellBoltzmann distribution (which is strictly only the case in idealized gases). Therefore, it is
a reasonable choice to initialize the atoms velocities according to the MaxwellBoltzmann equation in the
beginning of a MD simulation. The goal is to find an initial velocity distribution in which each degree of
freedom possesses a similar amount of energy, such that the equipartition theorem is approximately fulfilled.
The scalar MaxwellBoltzmann velocity distribution (leaving out the normalization factor) at temperature T
is given by
mv 2
f (v) = v 2 exp . (9.315)
2kB T
To initialize the particles velocities such that this distribution function is fulfilled, one starts with a series of
normal-distributed random numbers with mean 0 and variance 1, denoted by N (0, 1). The cartesian velocity
components for each atom are then computed by
r
kB T
vi, := N (0, 1) , {x, y, z}, i = 1 . . . N. (9.316)
mi
As the C++98 standard does not offer a platform-independent way of obtaining normal-distributed random
numbers, these are internally computed from uniformly distributed random numbers by applying the Box
Muller transform: Assuming that u1 and u2 are two uniformly distributed random numbers from the interval
[0, 1], the equations
p
z1 := 2 log (u1 ) cos (2u2 ) , (9.317)
p
z2 := 2 log (u1 ) sin (2u2 ) (9.318)
yield two new random numbers z1 and z2 which obey a normal distribution with mean 0 and variance 1.
After the velocities have been initialized, the total linear momentum of the system will probably have some
finite value other than zero. As the linear momentum is (approximately) conserved within a molecular
dynamics simulation, this would result in the system drifting away into one direction during the course of the
simulation, which is probably not desired. Therefore, the total momentum is explicitly set to zero after the
MaxwellBoltzmann initialization:
N
X
P~tot := mi~vi,old , (9.319)
i=1
P~tot
~vi,new := ~vi,old , i = 1 . . . N. (9.320)
mi N
This, of course, might change the initial temperature. Therefore, a final step is performed, in which all
velocity vectors are multiplied with a factor that is determined such that the initial temperature exactly
matches the target value.
9.26.3.3 Thermostats
After the initial velocities have been initialized to some finite temperature, it might be assumed that one
can simply start the time integration of the dynamical system (equivalent to the NVE ensemble), and the
starting temperature would be approximately preserved. In a real system, however, there are (at least) two
reasons why the temperature will strongly deviate from the initial value already after a few steps. First, the
initial velocity distribution only considers the kinetic energy of the particles, but some amount of energy
will be exchanged with the potential energy contribution (e. g., bond stretching) immediately, altering the
temperature. Secondly, the numerical errors introduced due to the finite time step (and in case of ab initio
MD, also due to the approximate forces) will lead to a drift in energy and therefore in temperature. To
counter these effects, it is often desirable to have a temperature control during the course of the simulation
(which then runs in the NVT ensemble), which is called a thermostat.
There exist many different kinds of thermostats, ranging from simple expressions up to highly complex
dynamical systems on their own. But all of them share a common issue: If the thermostat is coupled only
weakly to the system, the temperature will change anyway. However, if the thermostat is coupled more
strongly to the system (i. e., intervenes stronger), then the dynamics of the simulation will change, no longer
resembling the undisturbed original dynamics which one wants to investigate. Therefore, it is always a
tradeoff between temperature stability and disturbed dynamics to decide how strong a thermostat should be
coupled to the system.
At the moment, only the simple Berendsen thermostat is implemented in the ORCA MD module. More
thermostats (e. g., the widely used NoseHoover thermostat) will follow in a future release.
Berendsen Thermostat
The Berendsen thermostat is similar to the simple velocity rescaling scheme, but enhanced by a time constant
to control the coupling strength. Let T0 be the desired target temperature and T the current temperature
of the system. Then the temperature gradient caused by the thermostat can be expressed as
dT T0 T
= . (9.321)
dt
Considering the fact that discrete time steps t are used, the correction factor for the velocities in each time
step is determined by r
t (T0 T )
f := 1 + (9.322)
T
The new velocities are then easily obtained as
~vi,new := f ~vi,old , i = 1 . . . N. (9.323)

9.27 Implicit Solvation Models 675
Lets consider some special cases. If = t, the whole temperature deviation from T0 is corrected immediately,
such that the temperature is always exactly kept at the target value. This is identical to simple velocity
rescaling (without any time constant), which is known to work poorly for most systems (a single harmonic
oscillator would, e. g., simply explode). With a larger time constant > T , the coupling strength is reduced,
leading to reasonable results. Typically, a value of in the range of 20 . . . 200 T will be applied. For
, the coupling strength goes to zero, such that the thermostat is no longer active. Values of < T
are not allowed.
From the formula, it becomes clear that a Berendsen thermostat will have no effect if the system has a
temperature of 0 K (or in the massive case: if the considered degree of freedom has 0 K), because it is based
on multiplying the velocities by a factor to modify the temperature. Therefore, this type of thermostat cant
be used to heat a system up starting from 0 K.
9.27 Implicit Solvation Models
Implicit solvation models play an important role in quantum chemistry. Without resorting to placing multiple
solvation shells of solvent molecules implicit solvent models are able to mimic the effect of a specific solvent
on the solute.
The implicit solvent models implemented in ORCA are
C-PCM [396] : The Conductor-like Continuum Polarization Model
SMD [397] : The Solvation Model based on Density
The keywords to control these solvent models have been kept very similar to each other, so it is easy to switch
between the three.
9.27.1 The Conductor-like Polarizable Continuum Model (C-PCM)
The conductor-like polarizable continuum model (C-PCM) is an implementation of the conductor-like apparent
surface charge methods. In these models the solute is placed in a cavity of roughly molecular shape. The
solvent reaction field is described by apparent polarization charges on the cavity surface, which are in turn
determined by the solute. The cavity in ORCA is constructed using the GEPOL [398400] algorithm, either
as solvent-excluding surface, or solvent-accessible surface. The ORCA C-PCM implementation closely follows
the C-PCM [396] paper. The molecular Hamiltonian is perturbed by the solvent:
=H
H 0 + V (9.324)
0 is the Hamiltonian of the isolated molecule, whereas V describes the solute solvent interactions.
where H
The SCF procedure leads to the variational minimization of the free energy of the solute, G:
D E 1D E
0
G = H + V (9.325)

2
Using the conductor-like boundary condition the electrostatic potential can be determined by
tesserae
X
V (~r) + Vqi (~r) = 0 (9.326)
i
where V and Vqi are the electrostatic potential due to the solute and to the polarization charges and ~r is a
point on the cavity surface. The vector of polarization charge can then be determined by
AQ = V (9.327)
wheree the vector V contains the electrostatic potential due to the solute on the tesserae. The elements of
the matrix A are r
4
Aii = 1.07 (9.328)
Si
1
Aij = (9.329)
|~ri ~rj |
in which Si is the area of tessera i. If the conductor-like screening model is used to represent a model with a
dielectric constant , the actual charges become
1
q= Q (9.330)

The free energy of solvation can be approximately calculated as [401]
GS = (E + GCD ) E 0 (9.331)
where E 0 is the total energy of the molecule in vacuum and GCD is the nonelectrostatic contribution
(dispersion and cavity formation terms). It can be calculated as [401]
GCD = 1.321 + 0.0067639A (9.332)
where A is the cavity surface area.
The C-PCM model can be used via
! CPCM(solvent)
where solvent can be one of

Solvent Name Dielectric Constant Refractive Index

Water 80.4 1.33
Acetonitrile 36.6 1.344
Acetone 20.7 1.359
Ammonia 22.4 1.33
Ethanol 24.3 1.361
Methanol 32.63 1.329
CH2Cl2 9.08 1.424
CCl4 2.24 1.466
DMF 38.3 1.430
DMSO 47.2 1.479
Pyridine 12.5 1.510
THF 7.25 1.407
Chloroform 4.9 1.45
Hexane 1.89 1.375
Toluene 2.4 1.497
The parameters can be more accurately defined using the %cpcm block input. The available options are as
follows
%cpcm epsilon 80.0 # Dielectric constant

refrac 1.0 # Refractive index
rsolv 1.3 # Solvent probe radius
rmin 0.5 # Minimal GEPOL sphere radius
fepstype cpcm # Epsilon function type: cpcm, cosmo
surfacetype gepol_ses # GEPOL surface: gepol_ses, gepol_sas
ndiv 5 # Maximum depth for recursive sphere generation
radius[N] 1.3 # Atomic radius for atomic number N in Angstrom
end
9.27.2 The Conductor-like Screening Solvation Model (COSMO)
The COSMO solvation model has been removed from ORCA v4.0.0 !!!
Please use the C-PCM solvation model in combination with the COSMO epsilon function if required!
As a short form to use the C-PCM model with the COSMO epsilon function, you can specify the solvent via
!CPCMC(solvent).
9.27.3 The SMD Solvation Model
The SMD solvation model has been proposed by the Cramer and Truhlar groups [397], and is based on the
quantum mechanical charge density of a solute molecule interacting with a continuum description of the
solvent. In the model the full solute electron density is used without defining partial atomic charges and
the solvent is not represented explicitly but rather as a dielectric medium with the surface tension at the
solutesolvent boundary. SMD is a universal solvation model, in the sense that it is applicable to any charged
or uncharged solute in any solvent or liquid medium for which a few key descriptors are known. In particular,
these descriptors are the dielectric constant, refractive index, bulk surface tension, and acidity and basicity
parameters. Neglecting the concentration contribution, the model separates the observable solvation free
energy into two main components,
GS = GENP + GCDS . (9.333)
In ORCA, the first component is the bulk electrostatic contribution arising from a self-consistent reaction
field treatment that involves the electrostatic interaction using the Conductor-like Screening Model (COSMO).
However, the radii are set to intrinsic atomic Coulomb radii. The second component, called the cavity-
dispersion solvent-structure (CDS) term, is the contribution resulting from short-range interactions between
the solute and solvent molecules in the first solvation shell. This contribution is a sum of terms that are
proportional (with geometry-dependent proportionality constants called atomic surface tensions) to the
solvent-accessible surface areas of the individual atoms of the solute. The CDS contribution to the free energy
of solvation is given by
atoms
X atoms
X
GCDS = k Ak (R, RZk + rs ) + [M] Ak (R, RZk + rs ), (9.334)
k k
where k and [M] are the atomic surface tension of atom k and the molecular surface tension, respectively,
and Ak is the solvent accessible surface area (SASA). The SASA depends on the geometry R, the set RZk of
all atomic van der Waals radii, and the solvent radius rs , which is added to each of the atomic van der Waals
radii. In the program Bondi radii are used for CDS contribution.
More details can be found in the original paper of Marenich et al. [397], which should be cited in publications
using results of SMD calculations.
For the calculation of the CDS contributions to the energy and gradient ORCA calls the external tool
otool smd. This external tool is based on the GESOL code, [402] which is redistributed with permission of C.
J. Cramer and D. G. Truhlar.
SMD can be employed in single point calculations and geometry optimizations, using single-determinant SCF
(HF and DFT) and CASSCF methods. In post-SCF methods the result is corrected in the reference wave
function. To use SMD the user must simply specify smd true in the %cpcm block and provide the name of
the solvent. This automatically sets a number of default SMD parameters (see the table below for a list
of parametrized solvents included in the SMD library). If required, the user can also manually specify the
solvent descriptors used in an SMD calculation.
!CPCM(water)
%cpcm smd true # turn on SMD
solvent "water" # specify the name of solvent from the list
soln # index of refraction at optical frequencies at 293 K
soln25 # index of refraction at optical frequencies at 298 K
sola # Abrahams hydrogen bond acidity
solb # Abrahams hydrogen bond basicity
solg # relative macroscopic surface tension
solc # aromaticity, fraction of non-hydrogenic solvent atoms

# that are aromatic carbon atoms
solh # electronegative halogenicity, fraction of non-hydrogenic
# solvent atoms that are F, Cl, or Br
end
Before SCF start, the program prints the SMD information.
------------------------------------------------------------------------------
CPCM SOLVATION MODEL
------------------------------------------------------------------------------
CPCM parameters:
Epsilon ... 80.4000
Refrac ... 1.3300
Rsolv ... 1.3000
Surface type ... GEPOL SES
Epsilon function type ... CPCM
Solvent: ... WATER
SMD-CDS solvent descriptors:
Soln ... 1.3328
Soln25 ... 1.3323
Sola ... 0.0000
Solb ... 0.0000
Solg ... 0.0000
Solc ... 0.0000
Solh ... 0.0000
Radii:
Radius for N used is 3.5716 Bohr (= 1.8900 Ang.)
Calculating surface ... done! ( 0.0s)
GEPOL surface points ... 98
GEPOL Volume ... 279.2288
GEPOL Surface-area ... 211.1858
Calculating surface distance matrix ... done! ( 0.0s)
Performing Cholesky decomposition & store ... done! ( 0.0s)
Overall time for CPCM initialization ... 0.0s
After the SCF is converged, the output file shows the SMD contribution to the total energy.
*****************************************************
* SUCCESS *
* SCF CONVERGED AFTER 14 CYCLES *
*****************************************************
Old exchange energy = -22.029826261 Eh

New exchange energy = -22.029833040 Eh
Exchange energy change after final integration = -0.000006779 Eh
Total energy after final integration = -1664.968094721 Eh
Final COS-X integration done in = 258.813 sec
1.76169954764045 -2.99192151761839 5.55983672704991
1.969904463244464E-003 0.167822602012468 1.173801645083608E-002
Total Energy after outlying charge correction = -1664.968881930 Eh
SMD CDS free energy correction energy : -1.58900 Kcal/mol

Total Energy after SMD CDS correction = -1664.970626953 Eh
----------------
TOTAL SCF ENERGY
----------------
CPCM(ediel) : -0.00985444 Eh -0.26815 eV

SMD CDS (Gcds) : -0.00253223 Eh -0.06891 eV
At present there are 179 solvents in the SMD library:
solvent short name solvent short name

1 1,1,1-TRICHLOROETHANE 91 CYCLOPENTANE
2 1,1,2-TRICHLOROETHANE 92 CYCLOPENTANOL
3 1,2,4-TRIMETHYLBENZENE 93 CYCLOPENTANONE
4 1,2-DIBROMOETHANE 94 DECALIN (CIS/TRANS MIXTURE)
5 1,2-DICHLOROETHANE 95 CIS-DECALIN
6 1,2-ETHANEDIOL 96 N-DECANE
7 1,4-DIOXANE 97 DIBROMOMETHANE
8 1-BROMO-2-METHYLPROPANE 98 DIBUTYLETHER
9 1-BROMOOCTANE 99 O-DICHLOROBENZENE
10 1-BROMOPENTANE 100 E-1,2-DICHLOROETHENE
11 1-BROMOPROPANE 101 Z-1,2-DICHLOROETHENE
12 1-BUTANOL 102 DICHLOROMETHANE
13 1-CHLOROHEXANE 103 DIETHYL ETHER
14 1-CHLOROPENTANE 104 DIETHYL SULFIDE
15 1-CHLOROPROPANE 105 DIETHYLAMINE
16 1-DECANOL 106 DIIODOMETHANE
17 1-FLUOROOCTANE 107 DIISOPROPYL ETHER
18 1-HEPTANOL 108 CIS-1,2-DIMETHYLCYCLOHEXANE
19 1-HEXANOL 109 DIMETHYL DISULFIDE
20 1-HEXENE 110 N,N-DIMETHYLACETAMIDE
21 1-HEXYNE 111 N,N-DIMETHYLFORMAMIDE DMF
22 1-IODOBUTANE 112 DIMETHYLSULFOXIDE DMSO
23 1-IODOHEXADECANE 113 DIPHENYLETHER
24 1-IODOPENTANE 114 DIPROPYLAMINE
25 1-IODOPROPANE 115 N-DODECANE
26 1-NITROPROPANE 116 ETHANETHIOL
27 1-NONANOL 117 ETHANOL
28 1-OCTANOL 118 ETHYL ETHANOATE
29 1-PENTANOL 119 ETHYL METHANOATE
30 1-PENTENE 120 ETHYL PHENYL ETHER
31 1-PROPANOL 121 ETHYLBENZENE
32 2,2,2-TRIFLUOROETHANOL 122 FLUOROBENZENE
33 2,2,4-TRIMETHYLPENTANE 123 FORMAMIDE
34 2,4-DIMETHYLPENTANE 124 FORMIC ACID
35 2,4-DIMETHYLPYRIDINE 125 N-HEPTANE
36 2,6-DIMETHYLPYRIDINE 126 N-HEXADECANE
37 2-BROMOPROPANE 127 N-HEXANE
38 2-BUTANOL 128 HEXANOIC ACID
39 2-CHLOROBUTANE 129 IODOBENZENE
40 2-HEPTANONE 130 IODOETHANE
41 2-HEXANONE 131 IODOMETHANE
42 2-METHOXYETHANOL 132 ISOPROPYLBENZENE
43 2-METHYL-1-PROPANOL 133 P-ISOPROPYLTOLUENE
44 2-METHYL-2-PROPANOL 134 MESITYLENE
45 2-METHYLPENTANE 135 METHANOL
46 2-METHYLPYRIDINE 136 METHYL BENZOATE
47 2-NITROPROPANE 137 METHYL BUTANOATE
48 2-OCTANONE 138 METHYL ETHANOATE
49 2-PENTANONE 139 METHYL METHANOATE
50 2-PROPANOL 140 METHYL PROPANOATE
51 2-PROPEN-1-OL 141 N-METHYLANILINE
52 E-2-PENTENE 142 METHYLCYCLOHEXANE
53 3-METHYLPYRIDINE 143 N-METHYLFORMAMIDE (E/Z MIXTURE)
54 3-PENTANONE 144 NITROBENZENE PhNO2
55 4-HEPTANONE 145 NITROETHANE
56 4-METHYL-2-PENTANONE 146 NITROMETHANE MeNO2
57 4-METHYLPYRIDINE 147 O-NITROTOLUENE
58 5-NONANONE 148 N-NONANE
59 ACETIC ACID 149 N-OCTANE

60 ACETONE 150 N-PENTADECANE
61 ACETONITRILE MeCN 151 PENTANAL
62 ACETOPHENONE 152 N-PENTANE
63 ANILINE 153 PENTANOIC ACID
64 ANISOLE 154 PENTYL ETHANOATE
65 BENZALDEHYDE 155 PENTYLAMINE
66 BENZENE 156 PERFLUOROBENZENE
67 BENZONITRILE 157 PROPANAL
68 BENZYL ALCOHOL 158 PROPANOIC ACID
69 BROMOBENZENE 159 PROPANONITRILE
70 BROMOETHANE 160 PROPYL ETHANOATE
71 BROMOFORM 161 PROPYLAMINE
72 BUTANAL 162 PYRIDINE
73 BUTANOIC ACID 163 TETRACHLOROETHENE
74 BUTANONE 164 TETRAHYDROFURAN THF
75 BUTANONITRILE 165 TETRAHYDROTHIOPHENE-S,S-DIOXIDE
76 BUTYL ETHANOATE 166 TETRALIN
77 BUTYLAMINE 167 THIOPHENE
78 N-BUTYLBENZENE 168 THIOPHENOL
79 SEC-BUTYLBENZENE 169 TOLUENE
80 TERT-BUTYLBENZENE 170 TRANS-DECALIN
81 CARBON DISULFIDE 171 TRIBUTYLPHOSPHATE
82 CARBON TETRACHLORIDE CCl4 172 TRICHLOROETHENE
83 CHLOROBENZENE 173 TRIETHYLAMINE
84 CHLOROFORM 174 N-UNDECANE
85 A-CHLOROTOLUENE 175 WATER
86 O-CHLOROTOLUENE 176 XYLENE (MIXTURE)
87 M-CRESOL 177 M-XYLENE
88 O-CRESOL 178 O-XYLENE
89 CYCLOHEXANE 179 P-XYLENE
90 CYCLOHEXANONE
9.28 Calculation of Properties
9.28.1 Electric Properties
For the calculation of first order (electric dipole and quadrupole moments) and second order (polarizabilities)
electric properties, the %elprop module was implemented. The second order properties can be calculated
through the solution of the CP-SCF equations. Details are shown below:
%elprop
Dipole true
Quadrupole true
Polar 1 # analytic calculation
2 # numeric differentiation of dipole
# moments
3 # fully numeric second derivatives
Solver CG # CG(conjugate gradient)
# other options: DIIS or POPLE(default)
MaxDIIS 5 # max. dimension of DIIS method
Shift 0.2 # level shift used in DIIS solver
Tol 1e-8 # Convergence of the CP-SCF equations
# (norm of the residual)
MaxIter 64 # max. number of iterations in CPSCF
PrintLevel 2
EField 1e-4 # electric field step (a.u.) used in the

# numeric calculation of the polarizabilities
Origin 1 # Center of mass
2 # Center of nuclear charge
end
The most efficient and accurate way to calculate the polarizability analytically is to use the coupled-perturbed
SCF method. The most time consuming and least accurate way is the numerical second derivative of the
total energy. Note that the numerical differentiation requires: (a) tightly or even very tightly converged SCF
calculations and (b) carefully chosen field increments. If the field increment is too large then the truncation
error will be large and the values will be unreliable. On the other hand, if the field increment is too small the
numerical error associated with the finite difference differentiation will get unacceptably large up to the point
where the whole calculation becomes useless.
9.28.2 The Spin-Orbit Coupling Operator
Several variants of spin-orbit coupling operators can be used for property calculations [403]. They are based
on effective potential and mean-field approaches, and have various parameters that can be selected via the
%rel block.
%rel
# ---------------------------------------------------
# SPIN ORBIT COUPLING OPERATORS
# ---------------------------------------------------
SOCType 0 # none
1 # effective nuclear charge
2 # mean-field with atomic densities read from
# disk; similar to SOCType=4
3 # mean-field/effective potential (default)
4 # mean-field with atomic densities generated
# on the fly; see bellow
# ---------------------------------------------------
# Flags for construction of potential; operative
# only for SOCType 3.
# ---------------------------------------------------
SOCFlags 1,2,3,0 # (default)
# Flag 1 = 0 - do not include 1-electron terms
# = 1 - do include 1-electron terms
# Flag 2 = 0 - do not include Coulomb terms
# = 1 - compute Coulomb terms fully numeric
# = 2 - compute Coulomb term seminumeric
# = 3 - compute Coulomb term with RI approx
# = 4 - compute Coulomb term exactly
# Flag 3 = 0 - do not include exchange terms

# = 1 - do include local X-alpha exchange
# the X-Alpha parameter can be chosen via
# % rel Xalpha 0.7 (default)
# = 2 - same as 1 but with sign reversed
# = 3 - exchange via one-center exact
# integrals including the spin-other
# orbit interaction
# = 4 - all exchange terms full analytic
# (this is expensive)
# Flag 4 = 0 - do not include DFT local correlation
# terms
# = 1 - do include local DFT correlation (here
# this is done with VWN5)
#
SOCMaxCenter 4 # max. number of centers to include in
# the integrals (not fully consistently
# implemented yet; better leave equal to 4)
# Simple input equivalents that are described in [403]:
# SOMF(1X) = SOCType 3, SOCFlags 1,2,3,0 and SOCMaxCenter 4
# RI-SOMF(1X) = SOCType 3, SOCFlags 1,3,3,0 and SOCMaxCenter 4
# VEFF-SOC = SOCType 3, SOCFlags 1,3,1,1 and SOCMaxCenter 4
# VEFF(-2X)-SOC = SOCType 3, SOCFlags 1,3,2,1 and SOCMaxCenter 4
# AMFI = SOCType 3, SOCFlags 1,4,3,0 and SOCMaxCenter 1
# AMFI-A = SOCType 4, SOCFlags 1,4,3,0 and SOCMaxCenter 1
# (AMFI-like approach that uses pre-calculated
# atomic densities as described in [355];
# this can be combined with the SOCOff option to
# exclude contributions from specific atoms)
# -----------------------------------------------
# For the effective nuclear charge SOC operator
# the nuclear charges can be adjusted.
# -----------------------------------------------
Zeff[26] 0.0 # set the effective nuclear charge
# of iron (Z = 26) to zero
# -----------------------------------------------
# Neglecting SOC contributions from particular
# atoms
# -----------------------------------------------
SOCOff 0,5 # turn off the SOC for atoms 0 and 5
# this makes sense if the SOC operator
# has only one center contributions
# (e.g. effective nuclear charge)
The Breit-Pauli spin-orbit coupling operator is given by:

(1) + H
SOC = H
H (2) (9.335)
SOC SOC
with the one- and two-electron contributions
2 XX
(1) = (ri RA pi ) 2 X X 3
HSOC ZA 3 si ZA riA liA
si (9.336)
2 i |ri RA | 2 i
A A
2 X X (ri rj pi )
(2) =
H (
si + 2
sj ) (9.337)
SOC 3
2 i |ri rj |
j6=i
2 X X 3
lij rij (
si + 2
sj ) (9.338)
2 i
j6=i
This operator would be hard to handle exactly; therefore it is common to introduce mean field and/or effective
potential approaches in which the operator is written as an effective one-electron operator:
SOC (eff)
X
H = h i si (9.339)
i
The simplest approximation is to simply use the the one-electron part and regard the nuclear charges as
adjustable parameters. Reducing their values from the exact nuclear charge is supposed to account in
an average way for the screening of the nuclear charge by the electrons. In our code we use the effective
nuclear charges of Koseki et al. This approximation introduces errors which are usually smaller than 10%
but sometimes are larger and may approach 20% in some cases. The approximation is best for first row
main group elements and the first transition row (2p and 3d elements). For heavier elements it becomes
unreliable.
A much better approximation is to take the two-electron terms into account precisely. Without going into
details here the situation is as in Hartree-Fock (or density functional) theory and one gets Coulomb,
exchange and correlation terms. The correlation terms (evaluated in a local DFT fashion) are negligible and
can be safely neglected. They are optionally included and are not expensive computationally. The Coulomb
terms is (after the one-electron term) the second largest contribution and is expensive to evaluate exactly.
The situation is such that in the Coulomb-part the spin-other orbit interaction (the second term in the
two-electron part) does not contribute and one only has to deal with the spin-own-orbit contribution. The
exact evaluation is usually too expensive to evaluate. The RI and seminumeric approximation are much
more efficient and introduce only minimal errors (on the order of usually not more than 1 ppm in g-tensor
calculations for example) and are therefore recommended. The RI approximation is computationally more
efficient. The one-center approximation to the Coulomb term introduces much larger errors. The fully
numeric method is both slower and less accurate and is not recommended.
The exchange term has contributions from both the spin-own-orbit and spin-other-orbit interaction. These
are taken both into account in the mean-field approximation which is accessed by Flag 3 = 3. Here a
one-center approximation is much better than for the Coulomb term since both the integrals and the density
matrix elements are short ranged. Together with the Coulomb term this gives a very accurate SOC operator
which is recommended. The DFT-Veff operator suffers from not treating the spin-other-orbit part in the
exchange which gives significant errors (also, local DFT underestimates the exchange contributions from the
spin-same-orbit interaction by some 10% relative to HF but this is not a major source of error). However, it
is interesting to observe that in the precise analytical evaluation, the spin-other-orbit interaction is exactly -2
times the spin-own-orbit interation. Thus, in the DFT framework one gets a much better SOC operator if
the sign of the DFT exchange term is simply reversed! This is accessed by Flag 3 = 2.
Altogether the new SOC operators are a significant step forward and are also more accurate than those used
in other programs which either neglect some contributions or make more approximations. For compatibility
reasons the default is still the old effective nuclear charge model and therefore the SOC operator actually
wanted should be directly specified.
9.28.2.1 Exclusion of Atomic Centers
In ORCA it is possible to change the spin-orbit coupling operator in order to exclude contributions from
user-defined atoms. This approach can be useful, for example, in quantifying the contribution of the ligands
to the zero-field splitting (ZFS); for an application of this method see Ref. [404].
This is illustrated for the calculation of the SOC contribution to the ZFS of the triplet oxygen molecule.
Using the input below we start by a normal calculation of the ZFS, including both oxygen atoms. Note that
we use here the effective nuclear charge operator. This is required as not all implemented SOC operators are
compatible with the decomposition in terms of individual centers contributions.
! def2-TZVP def2-TZVP/c
%casscf nel 8
norb 6
mult 3,1
nroots 1,3
rel dosoc true
end
end
%rel
SOCType 1
end
*xyz 0 3
O 0 0 0
O 0 0 1.207
*
The calculated value of the D parameter is approximately 2.574 cm1 . In a second calculation we exclude
the contribution from the first oxygen atom. For this we change the %rel block to the one below.
%rel
SOCType 1
SOCOff 0
end
Now the D parameter is calculated to be approximately 0.643 cm1, a result that deviates quite significantly
from half of the value calculated previously, implying that non-additive effects are important. In addition to
the effective nuclear charge operator, the AMFI-A operator described previously can be used. Given that
this is based on precalculated atomic densities, it might be preferred for heavier elements where the effective
nuclear charge operator becomes unreliable. The method is not limited to CASSCF calculations as described
above, and can be used in DFT, MRCI and ROCIS calculations.
9.28.3 The EPR/NMR Module
Starting from release 2.4.30 the module orca eprnmr replaces the earlier separate modules for calculating
g-tensors and hyperfine couplings (orca dftg and orca hf). The use of the EPR/NMR module is rather
simple. The individual flags are given below.
%eprnmr
# Calculate the g-tensor using CP-KS theory
GTensor 1
# Calculate the D-tensor

DTensor so # spin orbit part
ss # spin-spin part
ssandso # both parts
DSOC qro # quasi-restricted method; must be done with the keyword !uno
pk # Pederson-Khanna method.
# NOTE: both approximations are only valid for
# pure functionals (no HF exchange)
cp # coupled-perturbed method (default)
cvw # van W
ullen method
DSS direct # directly use the canonical orbitals for the spin density
uno # use spin density from UNOs
PrintLevel n # Amount of output (default 2)
# For the solution of the CP-SCF equations:

Solver Pople # Pople solver (default)
CG # conjugate gradient
DIIS # DIIS type solver
MaxIter 32 # maximum number of iterations
MaxDIIS 12 # max. number of DIIS vectors (only DIIS)
Tol 1e-6 # convergence tolerance

LevelShift 0.2 # levelShift for DIIS (ignored otherwise)
Ori CenterOfMass
CenterOfNucCharge
CenterOfElCharge # (default)
OwnNuc # for NMR chemical shifts take
# the own nucleus as gauge
# origin; this removes the
# origin dependence but the
# basis set convergence may
# still be very slow
IGLO # IGLO treatment; only rigorous if there
# is no HF exchange
GIAO # not yet implemented
N # number of the atom to put the origin
X,y,z # explicit position of the origin
# For GIAO calculations of the NMR chemical shieldings, to use

# in conjunction with the keyword NMR in the first line :
# treatment of 1electron integrals in the RHS of the CPSCF equations

giao_1el = giao_1e_analytic # analytical, default
giao_1e_numeric # numerical - for testing only
# treatment of 2electron integrals in the RHS of the CPSCF equations

# various options for combination of approximations in coulomb (J) and
# exchange (K) part for HF and DFT. RIJK is default, all other should
# be used with caution and for testing only.
giao_2el = giao_2el_rijk # RIJK approximation, default
giao_2el_same_as_scf # use same scheme as in SCF
giao_2el_analytic # fully analytical, for testing only
giao_2el_rijonx # RIJ approximation with analytical K
giao_2el_cosjonx # COSJ approximation with analytical K
giao_2el_rijcosx # RIJ approximation with COSX approximation
giao_2el_cosjx # COSJ approximation with COSX apprxomation
giao_2el_exactjcosx # exact J with COSX approximation
giao_2el_exactjrik # exact J with RIK approximation
Nuclei = all type { flags }

# Calculate nuclear properties. Here the properties
# for all nuclei of type are calculated (type is
# an element name, for example Cu.)
# Flags can be the following:
# aiso - calculate the isotropic part of the HFC
# adip -
calculate the dipolar part of the HFC
# aorb -
2nd order contribution to the HFC from SOC
# fgrad -
calculate the electric field gradient
# rho -
calculate the electron density at the
# nucleus
# shift - NMR chemical shift tensor
# For example:
# calculates the hyperfine coupling for all nitrogen atoms
Nuclei = all N { aiso, adip, fgrad, rho};
# In addition you can change several parameters:

Nuclei = all N { PPP=39.1, QQQ=0.5, III=1.0 };
# PPP : the proportionality factor for this nucleus
# = ge*gN*betaE*betaN
# QQQ : the quadrupole moment for this nucleus
# III : the spin for this nucleus
# You can also calculate the hfcs for individual atoms

# as in the following example:
Nuclei = 1,5,8 { aiso, adip};
# NOTE: Counting starts with atom 1!

# CAUTION: All the nuclei, mentioned in one line
as above will be assigned the same isotopic mass !
end
A little background information might be in order:
9.28.3.1 Hyperfine and Quadrupole Couplings
The hyperfine coupling has three contributions:
(a) The isotropic Fermi contact term that arises from the finite spin density on the nucleus under investigation.
It is calculated for nucleus N from:

4 1

~N

aiso (N ) = hSz i ge gN e N R (9.340)
3
Here, hSz i is the expectation value of the z-component of the total spin, ge and gN arethe electron and
nuclear g-factors and e and N are the electron and nuclear magnetons respectively. R ~ N is the spin
density at the nucleus. The proportionality factor PN = ge gN e N is commonly used and has the dimensions
MHz bohr3 in ORCA.
(b) The spin dipole part that arises from the magnetic dipole interaction of the magnetic nucleus with the
magnetic moment of the electron. It is also calculated as an expectation value over the spin density as:
X
5
Adip 2

(N ) = PN klkl k rN 3~rN ~rN rN l (9.341)
where is the spin-density matrix and ~rN is a vector of magnitude rN that points from the nucleus in
question to the electron ({} is the set of basis functions).
(c) The second order contribution that arises from spin-orbit coupling. Presently ORCA can calculate all these
contributions. The first two are calculated as simple expectation values of the appropriate operators over the
self-consistent spin density, but the second order contribution requires the solution of the coupled-perturbed
SCF equations and is consequently computationally more demanding. The contribution can be written:
1 X kl

Aorb k hSOC

(N ) = PN
l (9.342)
2S I
kl
The derivative of the spin density is computed from solving the coupled-perturbed SCF equations with respect
to the nucleus-orbit coupling as perturbation. The nucleus-orbit coupling is represented by the operator
(A)
X
3
hNOC
(A) = riA li, (9.343)
i
where the sum is over electrons and A is the nucleus in question.
The field gradient tensor is closely related to the dipole contribution to the hyperfine coupling. The main
differences are that the electron instead of the spin density enters its calculation and that it contains a nuclear
contribution due to the surrounding nuclei. It is calculated from
P
5
2
V (N ) = Pkl k rN 3~rN ~rN rN l
kl (9.344)
+
P ~ 5 3R
ZA R ~ AN R
~ AN R2
AN AN
A6=N
with ZA as the nuclear charge of nucleus A and R~ AN as a vector of magnitude RAN that points from nucleus
A to nucleus N . P is the first order density matrix.
NOTE:
Hyperfine and quadrupole couplings are properties where the standard basis sets that have been
designed for geometry optimization and the like may not be entirely satisfactory (especially for atoms
heavier than Ne). You should probably look into tailoring the basis set according to your needs. While
it is likely that a later release will provide one or two special basis sets for core-property calculations
at this time you have to make sure yourself that the basis set has enough flexibility in the core region,
for example by uncontracting core basis functions and adding s-primitives with large exponents (or
using the decontraction feature, section 9.3). If you add these tight functions and use DFT make
sure that the numerical integration is still satisfactory. Use the SpecialGrid feature to enlarge grids
for individual atoms without increasing the computational effort too drastically.
For very heavy nuclei you may not get satisfactory results because scalar relativistic effects may become
large. These are available at the ZORA and DKH levels in ORCA if the flag PictureChange is set
to true in the %rel block. The situation is such that for DKH the implementation is only valid for
the field gradient while the ZORA methodology is consistent throughout. The IORA method is also
available but the results will be approximate. In my experience some nice results have been obtained
with ZORA for HFCs and very nice results for EFGs with DKH. In these quasi-relativistic calculations
with DFT one has to be extremely cautious with numerical integration accuracy and values of intacc
of 9.0 and higher have been used for heavier (transition metal) nuclei. Much lower values just produce
noise!
Second order HFCs require the calculation of the spin-orbit coupling contributions which in turn requires the
calculation of the coupled perturbed SCF equations. These effects can be quite significant for heavier nuclei
and should definitely be included for transition metal complexes. The spin-orbit coupling treatment used is
the same as described under 9.28.2.
9.28.3.2 The g-Tensor
The EPR g-tensor is a property that can be calculated as a second derivative of the energy and it is
implemented as such in ORCA for the SCF methods, e.g. HF and DFT. At the SCF level four contributions
arise:
(SZ)
g = ge (9.345)
(RMC) 2 X D E
g = Pkl k T l (9.346)
S
k,l
* +
(DSO) 1 X X
g = Pkl k (rA ) [rA rO rA, rO, ] l (9.347)

2S
k,l A
X P

(PSO)
k hSOC
kl

g =
k (9.348)
B
k,l
Here, ge is the free-electron g-value (=2.002319...), S is the total spin, the fine structure constant, P is
the spin density matrix, {} is the basis set, T is the kinetic energy operator, (rA ) an approximate radial
operator, hSOC the spatial part of an effective one-electron spin-orbit operator and B is a component of the
magnetic field. The derivative of the spin-density matrix is calculated from coupled-perturbed SCF theory
with respect to a magnetic field perturbation (orbital Zeeman term).
Accuracy. g-tensor calculations at the SCF level are not highly demanding in terms of basis set size. Basis
sets that give reliable SCF results (at least valence double-zeta plus polarization) usually also give reliable
g-tensor results. For many molecules the Hartree-Fock approximation will give reasonable predictions. In a
number of cases, however, it breaks down completely. DFT is more robust in this respect and the number
of molecules where it fails is much smaller. Among the density functionals, the hybrid functionals seem to
be the most accurate. In my hands PBE0 is perhaps the best although PWP1 and B3LYP are not much
worse. The GGA functionals such as BP, PW91, BLYP or PBE are equally good for small radicals but are
significantly inferior to their hybrid counterparts for transition metal complexes.
Gauge dependence. Unfortunately, the g-tensor is a gauge dependent property, that is, the results depend
on where the origin is chosen within the molecule. Unless fully invariant procedures (such as IGLO or GIAO)
are used, this undesirable aspect is always present in the calculations. GIAO will be implemented in later
releases but is not yet available in ORCA. However, if the choice of gauge origin is not outrageously poor,
the gauge dependence is usually so small that it can be ignored for all practical purposes, especially if large
basis sets are used. ORCA gives you considerable freedom in the choice of gauge origin. It can either be the
center of mass, the center of nuclear charge, the center of electronic charge (recommended), a special atom or
a user-defined point in space. It is wise to check the sensitivity of the results with respect to the choice of
origin, especially when small g-shifts on the order of only a few hundred ppm are calculated. The IGLO
method should be a much better treatment of the origin dependence but there is presently little experience
with it. At least for NMR chemical shifts (in the absence of HF exchange) it is the method of choice.
Spin-orbit coupling operator. In previous versions of the code, the g-tensor module used the parameteri-
zation of Koseki et al. [405407] for the spin-orbit operator. This is expected to be a reasonable approximation
for the 2p and 3d elements and less satisfactory for heavier main group or transition metal containing systems.
Thus, the main target molecules with the simple operators are radicals made of light atoms and first row
transition metal complexes. More accurate SOC operators (at only moderately increased computational
cost) have now been implemented and are described in section 9.28.2. With these operators there are fewer
restrictions. However, for very heavy elements they will suffer from the shortcomings of the Breit-Pauli
approximation and future releases will modify these operators to take into account the ZORA or DKH
corrections to the SOC.
9.28.3.3 Zero-Field-Splitting
It is well known that the ZFS consists of a first order term arising from the direct spin-spin interaction [408]:
* +
X X 2
2 rij KL 3 (rij )K (rij )L

(SS) 1
DKL = 0SS 5 {2
szi szj sxi sxj syi syj } 0SS (9.349)
2 S (2S 1) i j6=i rij
(K,L =x,y,z). Here is the fine structure constant ( 1/137 in atomic units), rij is the electronic distance
vector with magnitude rij and si is the spin-vector operator for the ith electron. |0SSi is the exact ground
state eigenfunction of the Born-Oppenheimer Hamiltonian with total spin S and projection quantum number
MS = S. Since the spin-spin interaction is of first order, it presents no particular difficulties. The more
complicated contribution to the D-tensor arises from the spin-orbit interaction, which gives a second order
contribution. Under the assumption that the spin-orbit coupling (SOC) operator can to a good approximation
SOC = P h
be represented by an effective one-electron operator (H SOCsi ), ref [162] has derived the following
i i
sum-over-states (SOS) equations for the SOC contribution to the ZFS tensor:
* +* +
SOC(0) 1 X X
K;SOC
X
L;SOC

DKL = 2 1 SS
0 h si,0 b SS SS
b h si,0 0SS
(9.350)

b i i
S

b(Sb =S) i i
* +* +
SOC(1) 1 X X
K;SOC
X
L;SOC

1 SS S1S1 S1S1 SS
DKL = 0 hi si,+1 b b hi si,1 0

b
S (2S 1)

b(Sb =S1) i i
(9.351)
SOC(+1) 1
DKL = (S+1)(2S+1)

P 1
P K;SOC
SS
S+1S+1 P L;SOC
S+1S+1
SS (9.352)
b 0 hi si,1 b
b hi si,+1 0

b(Sb =S+1) i i
Here the one-electron spin-operator for electron i has been written in terms of spherical vector operator
components si,m with m = 0, 1 and b = Eb E0 is the excitation energy to the excited state multiplet
SS
b (all MS components are degenerate at the level of the BO Hamiltonian).
One attractive possibility is to represent the SOC by the spin-orbit mean-field (SOMF) method developed
by Hess et al., [409] widely used in the AMFI program by Schimmelpfennig [410] and discussed in detail
by Berning et al. [411] as well as in ref. 403. In terms of an (orthonormal) one-electron basis, the matrix
elements of the SOMF operator are:

1elSOC
hK;SOC
rs = p h K q
P SOC 3 SOC 3
SOC (9.353)
+ Prs pq gK rs 2 pr gK sq 2 sq gK pr
rs
and:
2 XX
1elSOC (ri ) =
h 3
ZA riA liA;k (9.354)
k
2 i
A
2 3
gkSOC (ri, rj ) = lij;k rij (9.355)
2
liA = ( ri RA ) p
i is the angular momentum of the ith electron relative to nucleus A. The vector

riA = ri RA of magnitude riA is the position of the ith electron relative to atom A. Likewise, the vector

rij = rj of magnitude rij is the position of the ith electron relative to electron j and lij = (
ri ri
rj ) p
i
is its angular momentum relative to this electron. P is the charge density matrix of the electron ground state
(Ppq = 0SS Eqp 0SS with Eqp = a+

+ +
p aq + ap aq where ap and aq are the usual Fermion creation and
annihilation operators).
9.28.3.4 General Treatment
The zero-field splitting (ZFS) is typically the leading term in the Spin-Hamiltonian (SH) for transition
metal complexes with a total ground state spin S>1/2 (for reviews and references see chapter 11). Its net
effect is to introduce a splitting of the 2S + 1 MS levels (which are exactly degenerate at the level of the
Born-Oppenheimer Hamiltonian), even in the absence of an external magnetic field. Thus, an analysis and
interpretation of the ZFS is imperative if the information content of the various physical methods that are
sensitive to ZFS effects.
In 2007, we have developed a procedure that makes the ZFS calculation compatible with the language of
analytic derivatives. [164] Perhaps the most transparent route is to start from the exact solutions of the
Born-Oppenheimer
D E Hamiltonian. To this end, we look at the second derivative of the ground state energy
SS SS
(E = 0 H 0 ) with respect to a spin-dependent one-electron operator of the general form:
K;(m) = x(m)
X
h hK (m)
pq Spq (9.356)
K
pq
Where hKpq is the matrix of the Kth component of the spatial part of the operator (assumed to be imaginary
(m)
Hermitian as is the case for the spatial components of the SOC operator) and Spq is the second quantized
(m)
form of the spin vector operator (m = 0, 1). The quantity xK is a formal perturbation parameter. Using
the exact eigenfunctions of the BO operator, the first derivative is:

E X
(m)
= hK (m)
pq Ppq (9.357)
x K (m)
xK =0 pq
With the components of the spin density:
D E
(m)
Ppq = 0SS |Spq
(m) SS
|0 (9.358)
The second derivative with respect to a second component for m0 = m is:

(m)
2E X Ppq
= hK (9.359)

(m) (m)
pq (m)
x x
K L (m)
xK =xL
(m)
=0 pq xL
The derivative of the spin density may be written as:
(m)
Ppq D
SS(m) (m) SS
E D
(m) SS(m)
E
(m)
= 0L |Spq |0 + 0SS |Spq |0L (9.360)
xL
Expanding the perturbed wavefunction in terms of the unperturbed states gives to first order:
E D E
SS(m) X L;(m) SS
0L = n 6= 01
n |ni n h 0 (9.361)
0 0E
Where |ni is any of the bS M . Thus, one gets:

(m)
2E X Ppq
(m) (m)
= hK
pq (m)
(9.362)
xK xL pq xL
X hD ED E D ED Ei
= 1 L;(m) |n
0SS |h n|h K;(m) |n n|h
K;(m) |0SS + 0SS |h L;(m) |0SS (9.363)
n
n6=0
The equality holds for exact states. For approximate electronic structure treatments, the analytic derivative
approach is more attractive since an infinite sum over states can never be performed in practice and the
calculation of analytic derivative is computationally less demanding than the calculation of excited many
electron states.
Using eq. 9.362, the components of the SOC-contribution to the D-tensor are reformulated as
(0)
SOC(0) 1 X K;SOC Ppq
DKL = 2
hpq (0)
(9.364)
2S pq x L
(+1)
SOC(1) 1 X Ppq
DKL = hK;SOC (1) (9.365)
S (2S 1) pq pq x L
(1)
SOC(+1) 1 X Ppq
DKL = hK;SOC
pq (+1)
(9.366)
(S + 1) (2S + 1) pq x L
These are general equations that can be applied together with any non-relativistic or scalar relativistic
electronic structure method that can be cast in second quantized form. Below, the formalism is applied to
the case of a self-consistent field (HF, DFT) reference state.
For DFT or HF ground states, the equations are further developed as follows:
The SCF energy is:
Z Z
1 (r1 ) (r2 ) 1 X
ESCF = VNN + Ph+ + P P (| ) + cDF EXC [ , ] (9.367)

dr1 dr2 aX
2 |r1 r2 | 2
Here VNN is the nuclear repulsion energy and h is a matrix element of the one-electron operator which
contains the kinetic energy and electron-nuclear attraction terms (habi denotes the trace of the matrix product
ab). As usual, the molecular spin-orbitals p are expanded in atom centered basis functions ( = , ):
X
p (r) = cp (r) (9.368)

with MO coefficients cp . The two-electron integrals are defined as:
Z Z
1
(| ) = (r1 ) (r1 ) r12 (r2 ) (r2 ) dr1 dr2 (9.369)
The mixing parameter aX controls the fraction of Hartree-Fock exchange and is of a semi-empirical nature.
EXC [ , ] represent the exchange-correlation energy. The parameter cDF is an overall scaling factor that
allows one to proceed from Hartree-Fock theory (aX = 1, cDF = 0) to pure DFT (aX = 0, cDF = 1) to hybrid
DFT (0 < aX < 1, cDF = 1). The orbitals satisfy the spin-unrestricted SCF equations:
X

F = h + P (| ) aX P (| ) + cDF (|VXC |) (9.370)

EXC
With VXC = (r)
and P = P + P being the total electron density. For the SOC perturbation it is
customary to regard the basis set as perturbation independent. In a spin-unrestricted treatment, the first
derivative is:
ESCF X X
i |hK sm |i + i |hK sm |i = 0

(m)
= (9.371)
xK i i
For the second derivative, the perturbed orbitals are required. However, in the presence of a spin-dependent
perturbation they can no longer be taken as pure spin-up or spin-down orbitals. With respect to the Lth
component of the perturbation for spin-component m, the orbitals are expanded as:
;(m)L (m);L (m);L

X X
i (r) = Ua i a (r) + Ua i a (r) (9.372)
a a
;(m)L (m);L (m);L

X X
i (r) = Ua i a (r) + Ua i a (r) (9.373)
a a
Since the matrix elements of the spin-vector operator components are purely real and the spatial part of the
SOC operator has purely imaginary matrix elements, it follows that the first order coefficients are purely
imaginary. The second derivative of the total SCF energy becomes:
2 ESCF P n (m);L (m);L o

(m) (m) = Ua i a |hK sm |i + Ua i i |hK sm |a
xK xL i a n
(m);L (m);L o
a |hK sm |i + Ua i i |hK sm |a
P
+ Ua i
i a n
(m);L (m);L o (9.374)
P
+ Ua i
i a n
(m);L (m);L o
P
+ Ua i
i a
Examination of the three cases m = 0, 1 leads to the following equations for the D-tensor components:
(0)
(0) 1 X P
|hK;SOC |

DKL = 2 (0)
(9.375)
4S x
L
(1)
(+1) 1 X P
DKL = |hK;SOC | (+1)
(9.376)
2 (S + 1) (2S + 1) x L
(+1)
(1) 1 X P
DKL = |hK;SOC | (1)
(9.377)
2S (2S 1) x L
Where a special form of the perturbed densities has been chosen. They are given in the atomic orbital basis
as:
(0)
P X (0);L
X (0);L
(0)
= Ua i c
i ca + Ua i ci ca (9.378)
xL i a i a
(+1)
P X (1);L
X (1);L
(1)
= Ua i c
i ca Ua i c
a ci (9.379)
xL i a i a
(1)
P X (+1);L
X (+1);L
(+1)
= Ua i ca c
i + Ua i ci c
a (9.380)
xL i a i a
The special form of the coupled perturbed equations are implemented in ORCA run as follows: The
perturbation is of the general form hK sm . The equations 9.375-9.380 and 9.381-9.386 below have been written
in such a way that the spin integration has been performed but that the spin-dependent factors have been
dropped from the right-hand sides and included in the prefactors of eqs. 9.375-9.377. The explicit forms of
the linear equations to be solved are:
m = 0:

(0) K(0) K(m)
X
(0) Ub j {(b i |a j ) (j i |a b )} = a |hK |i

a i Ua i + aX (9.381)
j b

(0) K(0) K(m)
X
(0) Ub j {(b i |a j ) (j i |a b )} = a |hK |i

a i Ua i + aX (9.382)
j b
m = +1:

(0) K(+1) K(+1) K(+1)
X X
(0) (j i |a b ) = a |hK |i

a i Ua i + aX Ub j (b i |a j ) aX Ub j (9.383)
j b b j

(0) K(+1) K(+1) K(+1)
X X
(0)
a i Ua i + aX Ub j (b i |ab j ) aX Ub j (j i |a b ) = 0 (9.384)
j b b j
m = 1:

(0) K(1) K(1) K(1)
X X
(0) (j i |a b ) = a |hK |i

a i Ua i + aX Ub j (b i |ab j ) aX Ub j (9.385)
j b b j

(0) K(1) K(1) K(1)
X X
(0)
a i Ua i + aX Ub j (b i |a j ) aX Ub j (j i |a b ) = 0 (9.386)
j b b j
Note that these coupled-perturbed (CP) equations contain no contribution from the Coulomb potential or
any other local potential such as the exchange-correlation potential in DFT. Hence, in the absence of HF
exchange, the equations are trivially solved:

K(0) a |hK |i
Ua i = (0) (0)
(9.387)
a i

K(0) a |hK |i
Ua i = (0) (0)
(9.388)
a i

K(+1) a |hK |i
Ua i = (0) (0)
(9.389)
a i
K(+1)
Ua i =0 (9.390)

K(1) a |hK |i
Ua i = (0) (0)
(9.391)
a i
K(1)
Ua i =0 (9.392)
K(+1) K(1)
It is interesting that the reverse spin flip coefficients Ua i and Ua i are only nonzero in the presence
of HF exchange. In a perturbation expansion of the CP equations they arise at second order (V2 /2 ) while
the other coefficients are of first order (V/; V represents the matrix elements of the perturbation). Hence,
these contributions are of the order of 4 and one could conceive dropping them from the treatment in order
to stay consistently at the level of 2 . These terms were nevertheless kept in the present treatment.
Equations 9.381-9.386 are referred to as CP-SOC (coupled-perturbed spin-orbit coupling) equations. They
can be solved by standard techniques and represent the desired analogue of the CP-SCF magnetic response
equations solved for the determination of the g-tensor and discussed in detail earlier [202]. It is readily
confirmed that in the absence of HF exchange, eqs. 9.387-9.392 inserted into eqs. 9.375-9.380 lead back to a
modified Pederson-Khanna type treatment of the SOC contributions to the D-tensor [184]. In the framework
of the formalism developed above, the Pederson-Khanna formula can be re-written in the form:
(SOC;PK) P L;(0) L;(0)

i hK;SOC a Ua i + 4S1 2
1

i hK;SOC a Ua i
P
DKL = 4S 2
i ,a i ,a
L;(1) L;(+1) (9.393)
4S1 2 i hK;SOC a Ua i 4S1 2 i hK;SOC a Ua i
P P
i ,a i ,a
This equation was derived from second-order non-self-consistent perturbation theory without recourse to
spin-coupling. For the special case of no Hartree-Fock exchange, the main difference to the treatment
presented here is that the correct prefactors from eqs. 9.364-9.366 occur in front of the spin-flip contributions
rather than 1/(4S 2 ) in eq. 9.393. In the presence of HF exchange it is suggested that the consistent
generalization of the PK method are eqs. 9.375-9.377 with the 1/(4S 2 ) prefactors and this way the method
has been implemented as an option into the ORCA program.
For completeness, the evaluation of the spin-spin term in the SCF case proceeds conveniently through:
(SS) ge2 2 XX

5 2

DKL = P P P P r12 3r12,K r12,L KL r12 (9.394)
4 S (2S 1)
as derived by McWeeny and Mizuno and discussed in some detail by Sinnecker and Neese. [165] In this
reference it was found that DFT methods tend to overestimate the spin-spin contribution if the calculations
are based on a spin-unrestricted SCF treatment. A much better correlation with experiment was found for
open-shell spin restricted calculations. The origin of this effect proved to be difficult to understand but it
was shown in ref [181] that in the case of small spin-contamination, the results of ROKS calculations and of
those that are obtained on the basis of the spin-unrestricted natural orbital (UNO) determinant are virtually
indistinguishable. It is therefore optionally possible in the ORCA program to calculate the spin-spin term on
the basis of the UNO determinant.
9.28.4 Local Energy Decomposition
The DLPNO-CCSD(T) method provides very accurate correlation energies and allows to successfully predict
many chemical phenomena. In order to facilitate the interpretation of coupled cluster results, we have developed
the Local Energy Decomposition (LED) analysis [176], which permit to divide the interaction energy between
two fragments (including the Hartree-Fock reference) into physically meaningful contributions.
As a word of caution we emphasize that only the total energy is an observable and its decomposition is, to
some extent, arbitrary. Nevertheless, the LED analysis appears to be physically well grounded and logical
to us, it is straightforward to apply and comes at negligible computational cost. It makes no assumption
about the strength of the intermolecular interaction and hence it remains valid and consistent over the entire
potential energy surface. Alternative schemes, such as the popular symmetry adapted perturbation theory,
are perturbative in nature and hence are best applied to weakly interacting systems.
The idea of the LED analysis is rather simple. In local correlation methods occupied orbitals are localized
and can be readily assigned to pre-defined fragments in the molecule. The fragmentation is user defined.
The same can be done for the correlation energy in terms of pair correlation energies that refer to pairs of
occupied orbitals. Furthermore, since in the DLPNO-CCSD(T) method the virtual space is spanned by pair
natural orbitals that are also essentially local, the entire correlation energy can be decomposed into local
contributions as well.
The LED has as reference point the geometrically and electronically relaxed fragments that constitute the
interacting super-molecule. Relative to this reference state, the electronic energy difference to the interacting
9.29 Natural Bond Orbital (NBO) Analysis 699
geometric and electronic structure can be written as:
E =Egeoprep
HF C
+ Eelprep + Eelprep
(X,Y ) (X,Y )
+ Eelstat + Eexch
DISP (X,Y )
+ ECSP (9.395)
CT (XY ) CT (XY )
+ ECSP + ECSP
(X,Y ) (X,Y )
+ ECW P + EC(T ) .
Egeoprep is the energy needed to distort the fragments from their equilibrium configuration to the interacting
HF C
geometry. The Eelprep + Eelprep term represents the electronic preparation energy and describes how
much energy is necessary to bring the fragments into the electronic structure that is optimal for interaction. It
(X,Y )
includes both Hartree-Fock and correlation contributions. Eexch is the inter-fragment exchange interaction
(X,Y )
(it always gives a binding contribution in our formalism) and Eelstat is the electrostatic energy interaction.
These two terms are described at the Hatree-Fock level of theory. The decomposition of the DLPNO-CCSD
DISP (X,Y ) CT (XY ) CT (XY )
strong pairs yields ECSP , the genuine dispersion interaction term, and ECSP + ECSP , the
dynamic charge polarization terms that mainly serve to counteract the HF overpolarization. In a simplified
(X,Y ) (X,Y ) (X,Y )
scheme, this last contribution can be counted together with Eelstat . Finally, ECW P and EC(T ) are
(X,Y )
contributions from weak pairs and triple excitations, respectively. While ECW P is mainly of dispersive
DISP (X,Y )
nature and hence, in a simplified scheme, can be counted together with ECSP , the intermolecular
(X,Y )
contribution of the triples EC(T ) is expected to contain dispersion and charge transfer contributions as
well.
The following options can be used in accordance with LED.
! DLPNO-CCSD(T) cc-pVDZ cc-pVDZ/C cc-pVDZ/JK RIJK TightPNO LED TightSCF
%mdci
LED 1 # localization method for the PNOs. Choices:
# 1 = PipekMezey (default)
# 2 = FosterBoys
PrintLevel 3 # Selects large output for LED and prints the
# detailed contribution
# of each DLPNO-CCSD strong pair
LocMaxIterLed 300 # Maximum number of localization iterations for PNOs
LocTolLed 1e-6 # Absolute threshold for the localization procedure for PNOs
end
9.29 Natural Bond Orbital (NBO) Analysis
A popular and useful method for population analysis is the natural bond orbital analysis due to Weinhold
and co-workers. It is implemented in the NBO program which is distributed in older versions via the CCL
list and in newer versions via the University of Wisconsin/Madison. Information about the NBO program
can be found at http://www.chem.wisc.edu/nbo6. In order to use it together with ORCA you need a
version of the stand-alone executable. Starting with version 3.1.x ORCA can only be used with NBO6. To
specify the NBO executable the environment variable NBOEXE=/full/name/of/nbo6-executable has to be
set. As the NBO part of the interface is not independent of the integer data-type width (i4 or i8), the NBO
executable which will be used together with ORCA has to be compiled using i4!
ORCA features two methods to interface with the NBO program: ! NBO and the %nbo - block. The following
example illustrates the use:
#
# Test the interface to the NBO program
#
! RHF SVP NBO
* xyz 0 1
C 0.000000 0.000000 0.000000
O 1.200000 0.000000 0.000000
H -0.550000 0.952628 0.000000
H -0.550000 -0.952628 -0.000000
*
This produces the following output:
Now starting NBO6....
*********************************** NBO 6.0 ***********************************

N A T U R A L A T O M I C O R B I T A L A N D
N A T U R A L B O N D O R B I T A L A N A L Y S I S
***************************** development version *****************************
(c) Copyright 1996-2014 Board of Regents of the University of Wisconsin System
on behalf of the Theoretical Chemistry Institute. All rights reserved.
Cite this program as:
NBO 6.0. E. D. Glendening, J. K. Badenhoop, A. E. Reed,

J. E. Carpenter, J. A. Bohmann, C. M. Morales, C. R. Landis,
and F. Weinhold (Theoretical Chemistry Institute, University
of Wisconsin, Madison, WI, 2013); http://nbo6.chem.wisc.edu/
/NPA / : Natural Population Analysis

/NBO / : Natural Bond Orbital Analysis
/AONBO / : Checkpoint the AO to NBO transformation
/ARCHIVE/ : Write the archive file to lfn47
Filename set to FILE
Job title: ORCA Job: check
NATURAL POPULATIONS: Natural atomic orbital occupancies

NAO Atom No lang Type(AO) Occupancy Energy

---------------------------------------------------------
1 C 1 s Cor( 1s) 1.99966 -11.23369
2 C 1 s Val( 2s) 1.01121 -0.23403
3 C 1 s Ryd( 3s) 0.00689 0.61431
4 C 1 px Val( 2p) 0.82024 0.05722
5 C 1 px Ryd( 3p) 0.01280 0.63773
6 C 1 py Val( 2p) 1.09998 -0.01646
7 C 1 py Ryd( 3p) 0.00064 0.80092
8 C 1 pz Val( 2p) 0.65974 -0.03397
9 C 1 pz Ryd( 3p) 0.00302 0.62768
10 C 1 dxy Ryd( 3d) 0.00572 2.75504
11 C 1 dxz Ryd( 3d) 0.00375 2.25507
12 C 1 dyz Ryd( 3d) 0.00000 2.09062
13 C 1 dx2y2 Ryd( 3d) 0.00346 2.73607
14 C 1 dz2 Ryd( 3d) 0.00114 2.41304
15 O 2 s Cor( 1s) 1.99976 -20.31978

16 O 2 s Val( 2s) 1.70414 -1.12939
17 O 2 s Ryd( 3s) 0.00165 1.55528
18 O 2 px Val( 2p) 1.62356 -0.45319
19 O 2 px Ryd( 3p) 0.00083 1.29399
20 O 2 py Val( 2p) 1.91500 -0.46837
21 O 2 py Ryd( 3p) 0.00338 1.41525
22 O 2 pz Val( 2p) 1.32997 -0.28652
23 O 2 pz Ryd( 3p) 0.00010 1.30113
24 O 2 dxy Ryd( 3d) 0.00211 3.27368
25 O 2 dxz Ryd( 3d) 0.00339 3.20718
26 O 2 dyz Ryd( 3d) 0.00000 2.98898
27 O 2 dx2y2 Ryd( 3d) 0.00393 3.55727
28 O 2 dz2 Ryd( 3d) 0.00114 3.17796
29 H 3 s Val( 1s) 0.88637 0.05683

30 H 3 s Ryd( 2s) 0.00304 0.41662
31 H 3 px Ryd( 2p) 0.00031 2.18696
32 H 3 py Ryd( 2p) 0.00166 2.47708
33 H 3 pz Ryd( 2p) 0.00002 1.85384
34 H 4 s Val( 1s) 0.88637 0.05683

35 H 4 s Ryd( 2s) 0.00304 0.41662
36 H 4 px Ryd( 2p) 0.00031 2.18696
37 H 4 py Ryd( 2p) 0.00166 2.47708
38 H 4 pz Ryd( 2p) 0.00002 1.85384
Summary of Natural Population Analysis:
Natural Population
Natural ---------------------------------------------
Atom No Charge Core Valence Rydberg Total
--------------------------------------------------------------------
C 1 0.37176 1.99966 3.59117 0.03740 5.62824
O 2 -0.58895 1.99976 6.57267 0.01652 8.58895
H 3 0.10859 0.00000 0.88637 0.00503 0.89141
H 4 0.10859 0.00000 0.88637 0.00503 0.89141
====================================================================
* Total * 0.00000 3.99942 11.93659 0.06400 16.00000
Natural Population
---------------------------------------------------------
Core 3.99942 ( 99.9854% of 4)
Valence 11.93659 ( 99.4716% of 12)
Natural Minimal Basis 15.93600 ( 99.6000% of 16)
Natural Rydberg Basis 0.06400 ( 0.4000% of 16)
---------------------------------------------------------
Atom No Natural Electron Configuration

----------------------------------------------------------------------------
C 1 [core]2s( 1.01)2p( 2.58)3s( 0.01)3p( 0.02)3d( 0.01)
O 2 [core]2s( 1.70)2p( 4.87)3d( 0.01)
H 3 1s( 0.89)
H 4 1s( 0.89)
NATURAL BOND ORBITAL ANALYSIS:
Occupancies Lewis Structure Low High

Max Occ ------------------- ----------------- occ occ
Cycle Ctr Thresh Lewis non-Lewis CR BD nC LP (L) (NL)
============================================================================
1 2 1.90 15.89597 0.10403 2 4 0 2 0 0
----------------------------------------------------------------------------
Structure accepted: No low occupancy Lewis orbitals
-------------------------------------------------------
Core 3.99942 ( 99.985% of 4)
Valence Lewis 11.89656 ( 99.138% of 12)
================== =============================
Total Lewis 15.89597 ( 99.350% of 16)
-----------------------------------------------------
Valence non-Lewis 0.07868 ( 0.492% of 16)
Rydberg non-Lewis 0.02535 ( 0.158% of 16)
================== =============================
Total non-Lewis 0.10403 ( 0.650% of 16)
-------------------------------------------------------
(Occupancy) Bond orbital / Coefficients / Hybrids

------------------ Lewis ------------------------------------------------------
1. (1.99966) CR ( 1) C 1 s(100.00%)
1.0000 0.0000 0.0000 0.0000 0.0000
0.0000 0.0000 0.0000 0.0000 0.0000
0.0000 0.0000 0.0000 0.0000
2. (1.99976) CR ( 1) O 2 s(100.00%)
1.0000 0.0000 0.0000 0.0000 0.0000
0.0000 0.0000 0.0000 0.0000 0.0000
0.0000 0.0000 0.0000 0.0000
3. (1.98819) LP ( 1) O 2 s( 55.84%)p 0.79( 44.11%)d 0.00( 0.05%)
0.0000 0.7471 -0.0168 0.6641 0.0071
-0.0000 0.0000 0.0000 -0.0000 0.0000
-0.0000 0.0000 -0.0201 0.0099
4. (1.91727) LP ( 2) O 2 s( 0.00%)p 1.00( 99.89%)d 0.00( 0.11%)

0.0000 0.0000 0.0000 -0.0000 -0.0000
0.9994 -0.0103 0.0000 -0.0000 -0.0328
-0.0000 0.0000 0.0000 -0.0000
5. (1.99996) BD ( 1) C 1- O 2
( 33.33%) 0.5773* C 1 s( 0.00%)p 1.00( 99.44%)d 0.01( 0.56%)
0.0000 0.0000 0.0000 0.0000 -0.0000
-0.0000 0.0000 0.9949 -0.0673 -0.0000
0.0750 0.0000 -0.0000 0.0000
( 66.67%) 0.8165* O 2 s( 0.00%)p 1.00( 99.75%)d 0.00( 0.25%)
0.0000 -0.0000 0.0000 -0.0000 0.0000
-0.0000 0.0000 0.9987 -0.0085 0.0000
-0.0504 0.0000 0.0000 -0.0000
6. (1.99969) BD ( 2) C 1- O 2
( 32.65%) 0.5714* C 1 s( 32.16%)p 2.10( 67.37%)d 0.01( 0.47%)
0.0000 0.5628 -0.0695 0.8153 0.0946
0.0000 -0.0000 0.0000 -0.0000 -0.0000
0.0000 0.0000 0.0620 -0.0288
( 67.35%) 0.8207* O 2 s( 44.21%)p 1.25( 55.49%)d 0.01( 0.30%)
0.0000 0.6643 0.0279 -0.7446 -0.0216
-0.0000 0.0000 -0.0000 -0.0000 -0.0000
-0.0000 0.0000 0.0480 -0.0264
7. (1.99572) BD ( 1) C 1- H 3
( 56.61%) 0.7524* C 1 s( 33.99%)p 1.94( 65.86%)d 0.00( 0.15%)
0.0000 0.5827 0.0187 -0.3995 -0.0027
0.7064 -0.0073 0.0000 -0.0000 -0.0314
-0.0000 0.0000 -0.0182 -0.0152
( 43.39%) 0.6587* H 3 s( 99.79%)p 0.00( 0.21%)
0.9989 -0.0092 0.0187 -0.0422 0.0000
8. (1.99572) BD ( 1) C 1- H 4
( 56.61%) 0.7524* C 1 s( 33.99%)p 1.94( 65.86%)d 0.00( 0.15%)
0.0000 0.5827 0.0187 -0.3995 -0.0027
-0.7064 0.0073 -0.0000 0.0000 0.0314
0.0000 0.0000 -0.0182 -0.0152
( 43.39%) 0.6587* H 4 s( 99.79%)p 0.00( 0.21%)
0.9989 -0.0092 0.0187 0.0422 -0.0000
---------------- non-Lewis ----------------------------------------------------
9. (0.00000) BD*( 1) C 1- O 2
( 66.67%) 0.8165* C 1 s( 0.00%)p 1.00( 99.44%)d 0.01( 0.56%)
( 33.33%) -0.5773* O 2 s( 0.00%)p 1.00( 99.75%)d 0.00( 0.25%)
10. (0.00002) BD*( 2) C 1- O 2
( 67.35%) 0.8207* C 1 s( 32.16%)p 2.10( 67.37%)d 0.01( 0.47%)
( 32.65%) -0.5714* O 2 s( 44.21%)p 1.25( 55.49%)d 0.01( 0.30%)
11. (0.03933) BD*( 1) C 1- H 3
( 43.39%) 0.6587* C 1 s( 33.99%)p 1.94( 65.86%)d 0.00( 0.15%)
0.0000 -0.5827 -0.0187 0.3995 0.0027
-0.7064 0.0073 -0.0000 0.0000 0.0314
0.0000 0.0000 0.0182 0.0152
( 56.61%) -0.7524* H 3 s( 99.79%)p 0.00( 0.21%)
-0.9989 0.0092 -0.0187 0.0422 -0.0000
12. (0.03933) BD*( 1) C 1- H 4
( 43.39%) 0.6587* C 1 s( 33.99%)p 1.94( 65.86%)d 0.00( 0.15%)
0.0000 -0.5827 -0.0187 0.3995 0.0027
0.7064 -0.0073 0.0000 -0.0000 -0.0314
-0.0000 0.0000 0.0182 0.0152
( 56.61%) -0.7524* H 4 s( 99.79%)p 0.00( 0.21%)
-0.9989 0.0092 -0.0187 -0.0422 0.0000

13. (0.01023) RY ( 1) C 1 s( 28.97%)p 2.39( 69.11%)d 0.07( 1.92%)
0.0000 0.0555 0.5354 0.1145 -0.8234
-0.0000 -0.0000 -0.0000 0.0000 -0.0000
-0.0000 0.0000 -0.1197 0.0696
14. (0.00530) RY ( 2) C 1 s( 0.00%)p 1.00( 11.80%)d 7.47( 88.20%)
0.0000 0.0000 0.0000 0.0000 -0.0000
0.0452 0.3406 0.0000 -0.0000 0.9391
0.0000 0.0000 -0.0000 -0.0000
15. (0.00002) RY ( 3) C 1 s( 46.15%)p 0.23( 10.74%)d 0.93( 43.12%)
16. (0.00000) RY ( 4) C 1 s( 16.32%)p 0.89( 14.59%)d 4.23( 69.09%)
17. (0.00000) RY ( 5) C 1 s( 0.00%)p 1.00(100.00%)d 0.00( 0.00%)
18. (0.00000) RY ( 6) C 1 s( 0.00%)p 1.00( 88.39%)d 0.13( 11.61%)
19. (0.00000) RY ( 7) C 1 s( 0.00%)p 1.00( 0.56%)d99.99( 99.44%)
20. (0.00000) RY ( 8) C 1 s( 0.00%)p 0.00( 0.00%)d 1.00(100.00%)
21. (0.00000) RY ( 9) C 1 s( 8.42%)p 0.75( 6.28%)d10.13( 85.30%)
22. (0.00322) RY ( 1) O 2 s( 0.00%)p 1.00( 98.62%)d 0.01( 1.38%)
0.0000 0.0000 0.0000 -0.0000 -0.0000
0.0063 0.9931 -0.0000 -0.0000 -0.1175
0.0000 0.0000 0.0000 -0.0000
23. (0.00017) RY ( 2) O 2 s( 21.42%)p 2.91( 62.34%)d 0.76( 16.24%)
0.0000 -0.0179 0.4624 0.0539 -0.7877
0.0000 -0.0000 -0.0000 0.0000 0.0000
0.0000 0.0000 0.3540 -0.1926
24. (0.00001) RY ( 3) O 2 s( 0.62%)p38.98( 24.26%)d99.99( 75.12%)
25. (0.00000) RY ( 4) O 2 s( 76.20%)p 0.17( 13.08%)d 0.14( 10.73%)
26. (0.00000) RY ( 5) O 2 s( 0.00%)p 1.00(100.00%)d 0.00( 0.00%)
27. (0.00000) RY ( 6) O 2 s( 0.00%)p 1.00( 1.49%)d66.18( 98.51%)
28. (0.00000) RY ( 7) O 2 s( 0.00%)p 1.00( 0.25%)d99.99( 99.75%)
29. (0.00000) RY ( 8) O 2 s( 0.00%)p 0.00( 0.00%)d 1.00(100.00%)
30. (0.00000) RY ( 9) O 2 s( 1.71%)p 0.43( 0.73%)d57.04( 97.56%)
31. (0.00314) RY ( 1) H 3 s( 99.37%)p 0.01( 0.63%)
0.0099 0.9968 -0.0772 -0.0189 0.0000
32. (0.00003) RY ( 2) H 3 s( 0.20%)p99.99( 99.80%)
33. (0.00002) RY ( 3) H 3 s( 0.00%)p 1.00(100.00%)
34. (0.00001) RY ( 4) H 3 s( 0.64%)p99.99( 99.36%)
35. (0.00314) RY ( 1) H 4 s( 99.37%)p 0.01( 0.63%)
0.0099 0.9968 -0.0772 0.0189 0.0000
36. (0.00003) RY ( 2) H 4 s( 0.20%)p99.99( 99.80%)
37. (0.00002) RY ( 3) H 4 s( 0.00%)p 1.00(100.00%)
38. (0.00001) RY ( 4) H 4 s( 0.64%)p99.99( 99.36%)
NHO DIRECTIONALITY AND BOND BENDING (deviations from line of nuclear centers)
[Thresholds for printing: angular deviation > 1.0 degree]

p-character > 25.0%
orbital occupancy > 0.10e
Line of Centers Hybrid 1 Hybrid 2

--------------- ------------------- ------------------
NBO Theta Phi Theta Phi Dev Theta Phi Dev
===============================================================================
3. LP ( 1) O 2 -- -- 90.0 0.0 -- -- -- --
4. LP ( 2) O 2 -- -- 90.0 90.1 -- -- -- --
5. BD ( 1) C 1- O 2 90.0 0.0 0.0 0.0 90.0 0.0 0.0 90.0
SECOND ORDER PERTURBATION THEORY ANALYSIS OF FOCK MATRIX IN NBO BASIS
Threshold for printing: 0.50 kcal/mol

E(2) E(NL)-E(L) F(L,NL)
Donor (L) NBO Acceptor (NL) NBO kcal/mol a.u. a.u.
===============================================================================
within unit 1
1. CR ( 1) C 1 23. RY ( 2) O 2 0.56 12.79 0.076
1. CR ( 1) C 1 31. RY ( 1) H 3 0.53 11.66 0.070
1. CR ( 1) C 1 35. RY ( 1) H 4 0.53 11.66 0.070
2. CR ( 1) O 2 13. RY ( 1) C 1 6.44 20.90 0.328
3. LP ( 1) O 2 11. BD*( 1) C 1- H 3 0.56 1.63 0.027
3. LP ( 1) O 2 12. BD*( 1) C 1- H 4 0.56 1.63 0.027
3. LP ( 1) O 2 13. RY ( 1) C 1 12.74 1.52 0.124
4. LP ( 2) O 2 11. BD*( 1) C 1- H 3 26.31 1.16 0.156
4. LP ( 2) O 2 12. BD*( 1) C 1- H 4 26.31 1.16 0.156
4. LP ( 2) O 2 14. RY ( 2) C 1 5.71 3.02 0.117
4. LP ( 2) O 2 27. RY ( 6) O 2 0.72 3.76 0.047
6. BD ( 2) C 1- O 2 13. RY ( 1) C 1 1.25 1.99 0.045
7. BD ( 1) C 1- H 3 12. BD*( 1) C 1- H 4 0.68 1.45 0.028
7. BD ( 1) C 1- H 3 22. RY ( 1) O 2 1.95 2.16 0.058
8. BD ( 1) C 1- H 4 11. BD*( 1) C 1- H 3 0.68 1.45 0.028
8. BD ( 1) C 1- H 4 22. RY ( 1) O 2 1.95 2.16 0.058
NATURAL BOND ORBITALS (Summary):
Principal Delocalizations
NBO Occupancy Energy (geminal,vicinal,remote)
===============================================================================
Molecular unit 1 (CH2O)
------ Lewis --------------------------------------
1. CR ( 1) C 1 1.99966 -11.23369 23(v),31(v),35(v)
2. CR ( 1) O 2 1.99976 -20.31978 13(v)
3. LP ( 1) O 2 1.98819 -0.93621 13(v),11(v),12(v)
4. LP ( 2) O 2 1.91727 -0.46967 11(v),12(v),14(v),27(g)
5. BD ( 1) C 1- O 2 1.99996 -0.53513
6. BD ( 2) C 1- O 2 1.99969 -1.40029 13(g)
7. BD ( 1) C 1- H 3 1.99572 -0.75649 22(v),12(g)
8. BD ( 1) C 1- H 4 1.99572 -0.75649 22(v),11(g)
------ non-Lewis ----------------------------------
9. BD*( 1) C 1- O 2 0.00000 0.20694
10. BD*( 2) C 1- O 2 0.00002 0.94920
11. BD*( 1) C 1- H 3 0.03933 0.68958
12. BD*( 1) C 1- H 4 0.03933 0.68958
13. RY ( 1) C 1 0.01023 0.58478
14. RY ( 2) C 1 0.00530 2.55081
15. RY ( 3) C 1 0.00002 1.54517
16. RY ( 4) C 1 0.00000 2.10893
17. RY ( 5) C 1 0.00000 0.64516
18. RY ( 6) C 1 0.00000 0.99534
19. RY ( 7) C 1 0.00000 2.24380
20. RY ( 8) C 1 0.00000 2.09062
21. RY ( 9) C 1 0.00000 2.07758
22. RY ( 1) O 2 0.00322 1.40263

23. RY ( 2) O 2 0.00017 1.55897
24. RY ( 3) O 2 0.00001 3.28708
25. RY ( 4) O 2 0.00000 1.77823
26. RY ( 5) O 2 0.00000 1.30247
27. RY ( 6) O 2 0.00000 3.28761
28. RY ( 7) O 2 0.00000 3.20733
29. RY ( 8) O 2 0.00000 2.98898
30. RY ( 9) O 2 0.00000 2.95015
31. RY ( 1) H 3 0.00314 0.42667
32. RY ( 2) H 3 0.00003 2.56145
33. RY ( 3) H 3 0.00002 1.85384
34. RY ( 4) H 3 0.00001 2.07427
35. RY ( 1) H 4 0.00314 0.42667
36. RY ( 2) H 4 0.00003 2.56145
37. RY ( 3) H 4 0.00002 1.85384
38. RY ( 4) H 4 0.00001 2.07427
-------------------------------
Total Lewis 15.89597 ( 99.3498%)
Valence non-Lewis 0.07868 ( 0.4917%)
Rydberg non-Lewis 0.02535 ( 0.1584%)
-------------------------------
Total unit 1 16.00000 (100.0000%)
Charge unit 1 0.00000
$CHOOSE
LONE 2 2 END
BOND D 1 2 S 1 3 S 1 4 END
$END
Maximum scratch memory used by NBO was 47171 words

Stopping NBO6...Storing NBOs: check.nbo
*** returned from NBO program ***
Thus, in this example the NBO analysis shows that a single Lewis structure is dominant with single bonds
between C and H, a double bond between C and O and two lone pairs at the oxygen just as ordinary
chemical arguments would imply. In addition, the program produces the four corresponding valence antibonds
(as expected). The remaining components of the basis set span the Rydberg space and lead to semilocalized,
orthogonal orbitals that are assigned to single atoms (Note the nomenclature: BD = bond, BD* = antibond,
LP = lone pair, CR = core orbital, RY= Rydberg orbital). The NPA analysis shows a patially negative
oxygen and partially positive carbon and hydrogen atoms all matching the expectations.
Additionally, the NBO orbitals are stored in the ORCA .gbw file-format in a file called jobname.nbo. This
file can be used for further analysis and usage with ORCA e.g. for plotting orbitals via orca plot.
The NBO program has many additional features and analysis tools. The features that are implemented in
ORCA are
%nbo
NBOKEYLIST = "$NBO ... $END"
DELKEYLIST = "$DEL ... $END"
COREKEYLIST = "$CORE ... $END"

NRTSTRKEYLIST = "$NRTSTR ... $END"
end
The syntax of the respective keylists is given by the NBO6.x manual.
Specifying the ! NBO keyword as a one-liner command, the corresponding block as NBOKEYLIST would be:
%nbo
NBOKEYLIST = "$NBO NBO NPA AONBO=C ARCHIVE $END"
end
The full set of features beyond those which can be give via the %nbo block can be accessed using the file
FILE.47, which is generated by the NBO program. This is an ascii file that can be edited with a text editor.
Add or remove keywords in the corresponding blocks as needed and call the gennbo program like
gennbo < FILE.47 >jobname.nboout
(a) (b) LP (c)
(d) (e) LP (f)
Figure 9.46: Six NBOs of the H2 CO molecule. Shown are the occupied bonding and orbitals
(left) for C and O, the two oxygen lone-pairs (middle) and the two and antibonding
orbitals (right).
The FILE.47 file looks like:

$GENNBO NATOMS=4 NBAS=38 UPPER BODM FORMAT $END

$NBO $END
$COORD
ORCA Job: check
6 6 0.000000 0.000000 0.000000
8 8 2.267671 0.000000 0.000000
1 1 -1.039349 1.800206 0.000000
1 1 -1.039349 -1.800206 0.000000
$END
$BASIS
If you have no need for this (rather large) file, then you have to delete it manually!
9.29.1 NBO Deletions
An advanced feature which has been implemented via the ORCA-NBO interface is the possibilty of using
deletions.
! RHF 3-21G BOHRS TightSCF
%nbo
nbokeylist="$nbo nbo npa aonbo=c archive $end"
delkeylist="$del lewis delete 1 element 3 11 $end"
end
*xyz 0 1
C 1.4089705283 0.0210567401 0.0000000000
N -1.3645072652 -0.1355759321 0.0000000000
H 1.9849776453 1.9986808971 0.0000000000
H 2.1492280974 -0.9096841007 1.6818209547
H 2.1492280974 -0.9096841007 -1.6818209547
H -2.0504340036 0.7268536543 -1.5583845544
H -2.0504340036 0.7268536543 1.5583845544
*
The DELKEYLIST provides NBO with the task to perform certain deletions of orbitals/interactions. Per
deletion ORCA calculates a new Fock-matrix on basis of an NBO-density corresponding to the deletions:
Starting NBO again for $del instructions...
LEWIS: Delete all non-Lewis NBOs

Deletion of the following orbitals from the NBO Fock matrix:
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
Orbital occupancies:
Orbital No deletions This deletion Change

------------------------------------------------------------------------------
1. CR ( 1) C 1 1.99900 2.00000 0.00100
2. CR ( 1) N 2 1.99953 2.00000 0.00047
3. LP ( 1) N 2 1.97796 2.00000 0.02204
4. BD ( 1) C 1- N 2 1.99855 2.00000 0.00145
5. BD ( 1) C 1- H 3 1.99863 2.00000 0.00137
6. BD ( 1) C 1- H 4 1.99400 2.00000 0.00600
7. BD ( 1) C 1- H 5 1.99400 2.00000 0.00600
8. BD ( 1) N 2- H 6 1.99441 2.00000 0.00559
9. BD ( 1) N 2- H 7 1.99441 2.00000 0.00559
10. BD*( 1) C 1- N 2 0.00012 0.00000 -0.00012
11. BD*( 1) C 1- H 3 0.01571 0.00000 -0.01571
12. BD*( 1) C 1- H 4 0.00771 0.00000 -0.00771
13. BD*( 1) C 1- H 5 0.00771 0.00000 -0.00771
14. BD*( 1) N 2- H 6 0.00428 0.00000 -0.00428
15. BD*( 1) N 2- H 7 0.00428 0.00000 -0.00428
16. RY ( 1) C 1 0.00103 0.00000 -0.00103
17. RY ( 2) C 1 0.00033 0.00000 -0.00033
18. RY ( 3) C 1 0.00023 0.00000 -0.00023
19. RY ( 4) C 1 0.00001 0.00000 -0.00001
20. RY ( 1) N 2 0.00110 0.00000 -0.00110
21. RY ( 2) N 2 0.00046 0.00000 -0.00046
22. RY ( 3) N 2 0.00038 0.00000 -0.00038
23. RY ( 4) N 2 0.00002 0.00000 -0.00002
24. RY ( 1) H 3 0.00178 0.00000 -0.00178
25. RY ( 1) H 4 0.00096 0.00000 -0.00096
26. RY ( 1) H 5 0.00096 0.00000 -0.00096
27. RY ( 1) H 6 0.00123 0.00000 -0.00123
28. RY ( 1) H 7 0.00123 0.00000 -0.00123
NEXT STEP: Perform one SCF cycle to evaluate the energy of the new density
matrix constructed from the deleted NBO Fock matrix.
------------------------------------------------------------------------------
Copying NBO density...
Calculating new Fock-Matrix...
Calculating Fock-Matrix...done!
New NBO energy via Fock-Matrix: -94.618095
Starting NBO again for $del/return energy instructions...
------------------------------------------------------------------------------
Energy of deletion : -94.618095196
Total SCF energy : -94.679444929
-------------------
Energy change : 0.061350 a.u., 38.498 kcal/mol
------------------------------------------------------------------------------
Multiple deletions can also be specified, as can be seen in the example above. The output then conains the
additional energy values:
Starting NBO again for $del instructions...
Deletion of the following NBO Fock matrix elements:

3, 11;
Orbital occupancies:
Orbital No deletions This deletion Change

------------------------------------------------------------------------------
1. CR ( 1) C 1 1.99900 1.99901 0.00001
2. CR ( 1) N 2 1.99953 1.99952 -0.00001
3. LP ( 1) N 2 1.97796 1.99343 0.01548
4. BD ( 1) C 1- N 2 1.99855 1.99867 0.00012
5. BD ( 1) C 1- H 3 1.99863 1.99850 -0.00014
6. BD ( 1) C 1- H 4 1.99400 1.99397 -0.00003
7. BD ( 1) C 1- H 5 1.99400 1.99397 -0.00003
8. BD ( 1) N 2- H 6 1.99441 1.99452 0.00011
9. BD ( 1) N 2- H 7 1.99441 1.99452 0.00011
10. BD*( 1) C 1- N 2 0.00012 0.00011 -0.00000
11. BD*( 1) C 1- H 3 0.01571 0.00049 -0.01522
12. BD*( 1) C 1- H 4 0.00771 0.00788 0.00017
13. BD*( 1) C 1- H 5 0.00771 0.00788 0.00017
14. BD*( 1) N 2- H 6 0.00428 0.00428 0.00000
15. BD*( 1) N 2- H 7 0.00428 0.00428 0.00000
16. RY ( 1) C 1 0.00103 0.00073 -0.00030
17. RY ( 2) C 1 0.00033 0.00033 0.00000
18. RY ( 3) C 1 0.00023 0.00036 0.00013
19. RY ( 4) C 1 0.00001 0.00001 -0.00000
20. RY ( 1) N 2 0.00110 0.00111 0.00001
21. RY ( 2) N 2 0.00046 0.00046 0.00000
22. RY ( 3) N 2 0.00038 0.00038 0.00000
23. RY ( 4) N 2 0.00002 0.00002 0.00000
24. RY ( 1) H 3 0.00178 0.00109 -0.00069
25. RY ( 1) H 4 0.00096 0.00100 0.00005
26. RY ( 1) H 5 0.00096 0.00100 0.00005
27. RY ( 1) H 6 0.00123 0.00123 0.00000
28. RY ( 1) H 7 0.00123 0.00123 0.00000
NEXT STEP: Perform one SCF cycle to evaluate the energy of the new density
matrix constructed from the deleted NBO Fock matrix.
------------------------------------------------------------------------------
Copying NBO density...
Calculating new Fock-Matrix...
Calculating Fock-Matrix...done!
New NBO energy via Fock-Matrix: -94.668195
Starting NBO again for $del/return energy instructions...
------------------------------------------------------------------------------
Energy of deletion : -94.668194723
Total SCF energy : -94.679444929
-------------------
Energy change : 0.011250 a.u., 7.060 kcal/mol
------------------------------------------------------------------------------
NOTE: Deletions are only implemented for SCF methods!
9.29.2 NBO for Post-HF Densities
NBO analysis can be performed on all methods producing a density. In some methods the density generation
has to be specified explictly, e. g. for MP2 calculations this would be:
! RHF MP2 3-21G TightSCF BOHRS NBO
%MP2
density relaxed
end
*xyz 0 1
C 1.4089705283 0.0210567401 0.0000000000
N -1.3645072652 -0.1355759321 0.0000000000
H 1.9849776453 1.9986808971 0.0000000000
H 2.1492280974 -0.9096841007 1.6818209547
H 2.1492280974 -0.9096841007 -1.6818209547
H -2.0504340036 0.7268536543 -1.5583845544
H -2.0504340036 0.7268536543 1.5583845544
*
The output will contain both the NBO analysis of the SCF density as well as of the MP2 relaxed density. An
NBO analysis of a density generated by the MDCI module can be specified as follows:
! RHF CISD 3-21G TightSCF BOHRS NBO
%mdci
density linearized
end
*xyz 0 1
C 1.4089705283 0.0210567401 0.0000000000
N -1.3645072652 -0.1355759321 0.0000000000
H 1.9849776453 1.9986808971 0.0000000000
H 2.1492280974 -0.9096841007 1.6818209547
H 2.1492280974 -0.9096841007 -1.6818209547
H -2.0504340036 0.7268536543 -1.5583845544
H -2.0504340036 0.7268536543 1.5583845544
*
Again, the output will contain both the NBO analysis of the SCF density as well es of the CISD linearized
density.
9.30 Population Analyses and Control of Output
At present ORCA knows three different ways of analyzing the computed SCF wavefunction that will be
described below. All of these methods can produce a tremendous amount of output. However, this output
can be precisely controlled by the user to his or her individual needs.
In general there is one compound key called PrintLevel which is there to choose reasonable amounts of
output. All that PrintLevel does is to set certain flags in the array Print which holds the details about
what to print and what not.
9.30.1 Controlling Output
The array Print allows the control of output. The general way of assigning elements of Print is:
%output
PrintLevel Normal
Print[ Flag ] 0 # turn print off
1 # turn print on
n # some flags are more sophisticated
end
The compound key PrintLevel can be used to select certain default settings for the print array. Specifying
Print after PrintLevel can be used to modify these defaults.
%output
PrintLevel Nothing
Mini
Small
Normal
Maxi
Large
Huge
Debug
end
Print has presently the following elements that can be user controlled:
Flag Action
P InputFile Echo the input file
P Cartesian Print the cartesian coordinates
9.30 Population Analyses and Control of Output 713
P Internal Print the internal coordinates

P Basis = 1 : Print the basis set information
= 2 : Also print the primitives in input format
P OneElec Print of the one electron matrix
P Overlap Print the overlap matrix
P KinEn Print the kinetic energy matrix
P S12 Print the S1/2 matrix
P GuessOrb Print the initial guess orbitals
P OrbEn Print Orbital Energies
P MOs Print the MO coefficients on convergence
P Density Print the converged electron density
P SpinDensity Print the converged spin density
P EHTDetails Print initial guess extended H uckel details
P SCFInfo Print the SCF input flags
P SCFIterInfo = 1 : print short iteration information
= 2 : print longer iteration information
= 3 : in a direct SCF also print integral progress
P Fockian Print Fockian matrix
P DIISMat Print DIIS matrix
P DIISError Print DIIS error
P Iter P Print Density
P Iter C Print MO coefficients
P Iter F Print Fock matrix
P Mayer Print Mayer population analysis. Default = on.
P NatPop Print Natural population analysis. Default = off.
P Hirshfeld Print Hirshfeld population analysis. Default = off.
P Mulliken Print Mulliken population analysis. Default = on
P AtCharges M Print Mulliken atomic charges
P OrbCharges M Print Mulliken orbital charges
P FragCharges M Print Mulliken fragment charges
P FragBondOrder M Print Mulliken fragment bond orders
P BondOrder M Print Mulliken bond orders
P ReducedOrbPop M Print Mulliken reduced orb. charges
P FragPopMO M Print Mulliken fragment population for each MO
P FragOvlMO M Print Mulliken overlap populations per fragment pair
P AtPopMO M Print Mulliken atomic charges in each MO
P OrbPopMO M Print Mulliken orbital population for each MO
P ReducedOrbPopMO M Print Mulliken reduced orbital population for each MO
P Loewdin Print Loewdin population analysis. Default = on.
P AtCharges L Print Loewdin atomic charges
P OrbCharges L Print Loewdin orbital charges
P FragCharges L Print Loewdin fragment charges
P FragBondOrder L Print Loewdin fragment bond orders
P BondOrder L Print Loewdin bond orders
P ReducedOrbPop L Print Loewdin reduced orb. charges

P FragPopMO L Print Loewdin fragment population for each MO
P FragOvlMO L Print Loewdin overlap populations per fragment pair
P AtPopMO L Print Loewdin atomic charges in each MO
P OrbPopMO L Print Loewdin orbital population for each MO
P ReducedOrbPopMO L Print Loewdin reduced orbital population for each MO
P NPA Natural population analysis
P NBO Natural bond orbital analysis
P Fragments Print fragment information
P GUESSPOP Print initial guess populations
P UNO FragPopMO M Print Mulliken fragment population per UNO
P UNO OrbPopMO M Print Mulliken orbital pop. per UNO
P UNO AtPopMO M Print Mulliken atomic charges per UNO
P UNO ReducedOrbPopMO M Print Mulliken reduced orbital pop. per UNO
P UNO FragPopMO L Print Loewdin fragment population per UNO
P UNO OrbPopMO L Print Loewdin orbital pop. per UNO
P UNO AtPopMO L Print Loewdin atomic charges per UNO
P UNO ReducedOrbPopMO L Print Loewdin reduced orbital pop. per UNO
P UNO OccNum Print occupation numbers per UNO
P AtomExpVal Print atomic expectation values
P AtomBasis Print atomic basis
P AtomDensFit Print electron density fit
P Symmetry Symmetry basic information
P Sym Salc Symmetry process printing
P SCFSTABANA Information on progress, convergence, and results of the
SCF stability analysis
P DFTD Print info on Grimmes dispersion correction
print mini = 0
print small = 1
print normal = 1
print maxi = 2
print huge = 2
P DFTD GRAD Print gradient info on Grimmes dispersion correction
print mini = 0
print small = 0
print normal = 0
print maxi = 1
print huge = 2
The various choices for PrintLevel have the following defaults:
PrintLevel Print settings

Mini P OrbEn = 1
P Cartesian = 1
P InputFile = 1
P SCFIterInfo = 1
Small all the previous plus
P SCFInfo = 1
P Mayer = 1
P MULLIKEN = 1
P AtCharges M = 1
P ReducedOrbPop M = 1
P Loewdin = 1
P AtCharges L = 1
P ReducedOrbPop L = 1
P Fragments = 1
P FragCharges M = 1
P FragBondOrder M = 1
P FragCharges L = 1
P FragBondOrder L = 1
Normal all the previous plus
P Internal = 1
P BondOrder L = 1
P BondOrder M = 1
P FragPopMO L = 1
P ReducedOrbPopMO L = 1
P SCFIterInfo = 2
Maxi all the previous plus
P GuessOrb = 1
P MOs = 1
P Density = 1
P SpinDensity = 1
P Basis = 1
P FragOVLMO M = 1
P OrbPopMO M = 1
P OrbCharges M = 1
Huge All the previous plus
P OneElec = 1
P Overlap = 1
P S12 = 1
P AtPopMO M = 1
P OrbPopMO M = 1
P AtPopMO L = 1
P EHTDetails = 1
Debug print everything
9.30.2 Mulliken Population Analysis
The Mulliken population analysis [412] is, despite all its known considerable weaknesses, the standard in
most quantum chemical programs. It partitions the total density using the assignment of basis functions to
given atoms in the molecules and the basis function overlap. If the total charge density is written as (~r)
and the total number of electrons is N we have:
Z
(~r) d~r = N (9.396)
and from the density matrix P and the basis functions {}:
X
(~r) = P (~r) (~r) (9.397)

therefore:
Z X Z
(~r) d~r = P (~r) (~r) d~r (9.398)
| {z }
S
X
= P S (9.399)

After assigning each basis function to a given center this can be rewritten:
XXX X
A B AB AB
= P S (9.400)
A B
XX X XXX X
A A AA AA A B AB AB
= P S + 2 P S (9.401)
A A B<A
Mulliken proposed to divide the second term equally between each pair of atoms involved and define the
number of electrons on center A, NA , as:
X X XX X
A A AA AA A B AB AB
NA = P S + P S (9.402)
B6=A
P
such that NA = N . The charge of an atom in the molecule is then:
A
QA = ZA NA (9.403)
where ZA is the core charge of atom A. The cross terms between pairs of basis functions centered on different
atoms is the overlap charge and is used in ORCA to define the Mulliken bond order:
X X
A B AB AB
BAB = 2 P S (9.404)

The Mulliken population analysis is turned on by using:
%output
Print[ P_Mulliken ] 1 # default = on
end
A number of additional options can be specified to control the details of the Mulliken population analysis.
By default the Mulliken population analysis is turned on.
%output
Print[ P_AtCharges_M ] 1 # Print atomic charges
Print[ P_OrbCharges_M ] 1 # Print orbital charges
Print[ P_FragCharges_M] 1 # Print fragment charges
Print[ P_BondOrder_M ] 1 # Print bond orders
Print[ P_FragBondOrder_M ] 1# Print fragment b.o.
Print[ P_ReducedOrbPop_M ] 1# Print reduced orb. Charges
Print[ P_AtPopMO_M ] 1 # Print atomic charges in
# each MO
Print[ P_OrbPopMO_M ] 1 # Print orbital populaiton
# for each MO
Print[ P_ReducedOrbPopMO_M] 1 # Print reduced orbital
# pop for each MO
Print[ P_FragPopMO_M ] 1 # Print the fragment
# population for for each MO
end
These options allow to get very detailed information about the computed wavefunctions and is much more
convenient than to look at the MOs directly. A reduced orbital population is a population per angular
momentum type. For example the sum of populations of each pz orbital at a given atom is the reduced
orbital population of the pz function.
Note that for finite temperature HF or KS-DFT calculations (SmearTemp > 0 K, fractional occupation
numbers or FOD analysis, see 9.5.8), only the Mulliken reduced orbital charges based on F OD will be printed.
They can be used to get a first impression about the localization of hot electrons in the molecule without
generating the corresponding FOD plot (see 9.31.2.2). The following example shows the corresponding
printout for the first carbon atom of p-benzyne based on a FOD analysis with default settings (see 9.5.8.2).
------------------------------------------
FOD BASED MULLIKEN REDUCED ORBITAL CHARGES
------------------------------------------
0 C s : 0.006371 s : 0.006371
pz : 0.016375 p : 0.030785
px : 0.009893
py : 0.004516
dz2 : 0.004248 d : 0.010308
dxz : 0.000254
dyz : 0.004855
dx2y2 : 0.000860
dxy : 0.000091
f0 : 0.000006 f : 0.000378
f+1 : 0.000014
f-1 : 0.000309
f+2 : 0.000002
f-2 : 0.000006
f+3 : 0.000010
f-3 : 0.000032
If other population analysis printouts are wanted the user is referred to the Lowdin analysis (9.30.3) which
is turned on by default using the total SCF density of the calculation, also in the case of finite electronic
temperature.
9.30.3 L
owdin Population Analysis
The Lowdin analysis [59] is somewhat more straightforward than the Mulliken analysis. In the Lowdin
method one changes to a basis where all overlap integrals vanish. This is accomplished via L owdins symmetric
1/2
orthogonalization matrix S . Using this transformation matrix the new basis functions are multicentered
but are in a least square sense as similar as possible to the original, strictly localized, atomic basis functions.
The similarity of the transformed functions and original functions is explored in the population analysis. The
density matrix transforms as:
PL = S1/2 PS1/2 (9.405)
Then the atomic populations are:
X
A L
NA = P (9.406)

The bond order is defined from the Wiberg index [413] that was first used in the context of semiempirical
methods (that are formulated in the L
owdin basis right from the start):
X X
A B L
2
BAB = P (9.407)

The output for the L

owdin population analysis (that I personally prefer over the Mulliken analysis) is closely
similar. By default the Lowdin population analysis is turned on and provides some more detail than the
Mulliken analysis.
%output
Print[ P_Loewdin ] 1 # default = on
end
The flags to regulate the details are almost identical:
%output
Print[ P_AtCharges_L ] 1 # Print atomic charges
Print[ P_OrbCharges_L ] 1 # Print orbital charges
Print[ P_FragCharges_L] 1 # Print fragment charges
Print[ P_BondOrder_L ] 1 # Print bond orders
Print[ P_FragBondOrder_L ] 1# Print fragment b.o.
Print[ P_ReducedOrbPop_L ] 1# Print reduced orb. Charges
Print[ P_AtPopMO_L ] 1 # Print atomic charges in
# each MO
Print[ P_OrbPopMO_L ] 1 # Print orbital population
# for each MO
Print[ P_ReducedOrbPopMO_L] 1 # Print reduced orbital
# pop for each MO
Print[ P_FragPopMO_L ] 1 # Print the fragment
# population for each MO
end
9.30.4 Mayer Population Analysis
Mayers bonding analysis [414417] is another creative attempt to define chemically useful indices. The Mayer
atomic charge is identical to the Mulliken charge. The Mayer bond order is defined as:
X X
A B
BAB = (PS) (PS) + (RS) (RS) (9.408)

Here P is the total electron density matrix and R is the spin-density matrix. These Mayer bond orders are
very useful. Mayers total valence for atom A is defined as:
X X
A A
VA = 2NA (PS) (PS) (9.409)

In normal bonding situations and with normal basis sets VA should be reasonably close to the valence of
atom A in a chemical sense (i.e. close to four for a carbon atom). The bonded valence is given by:
X
XA = VA BAB (9.410)
B6=A
and finally the free valence (a measure of the ability to form further bonds) is given by:
FA = VA XA (9.411)
The Mayer population analysis is turned on by:
%output
Print[ P_Mayer ] 1 # default = on
end
The output is rather simple and short and can not be further controlled. By default the Mayer population
analysis is turned on.
9.30.5 Natural Population Analysis
A popular and useful method for population analysis is the natural population analysis due to Weinhold and
co-workers. It is implemented in the NBO interface.
9.30.6 Density of the Effectively Unpaired Electrons
Investigating open-shell molecules, it is common practice to consult the spin-population analysis i.e. the
Mulliken spin-population using the spin-density matrix. The trace of spin-density corresponds to difference
in the number of and electrons. For an open-shell system in high-spin configuration, the trace matches
the number of unpaired electrons. However, the spin-density is not able to discriminate anti-ferromagnetic
coupled electrons from electron pairs. Thus, open-shell singlets appear to have no unpaired electrons. There
are many approaches in the literature to define a density matrix for unpaired electrons that correctly describe
low-spin configurations. For a CASSCF-type wavefunction, two common effective unpaired electron densities
(eff. UED) are available in ORCA. Namely, Taksukas [418] and Head-Gordons [419] formulations. If the
one-particle density matrix is denoted with , its eigenvalues n and vectors U satisfy
= UnU . (9.412)
The eff. unpaired electron density matrix UED can be expressed as a function of the natural orbital
occupation
UED = Uf (n)U . (9.413)
In case of Takatsukas definition we have

f (n) = 2n nn (9.414)
and for Head-Gordon the function is modified towards
f (n) = 1 |1 n|. (9.415)

As an example, let us recompute the example in [419] - two oxygen atoms at a 10 A distance in the triplet
state. The active space contains all the 2p orbitals and electrons. Since the ground-state is 9-fold degenerate,
we must take into account all 9 roots.
!svp
%maxcore 2000
%casscf nel 8
norb 6
mult 3
nroots 9
actorbs locorbs # localized active orbitals
end
%output
print[p_unpaired_t] 1 # Takatsuka UED
print[p_unpaired_hg] 1 # Head-Gordon UED
end
*xyz 0 3
O 0 0 0
O 0 0 10.0
*
According to the CASSCF wavefunction, there should be 4 unpaired electrons coupled to a triplet state.
---------------------------------------------
CAS-SCF STATES FOR BLOCK 1 MULT= 3 NROOTS= 9
---------------------------------------------
ROOT 0: E= -149.4274934720 Eh
0.49867 [ 31]: 121211
0.48600 [ 50]: 211112
0.01400 [ 51]: 211121
ROOT 1: E= -149.4274934720 Eh 0.000 eV
0.49867 [ 22]: 112211
0.48599 [ 51]: 211121
0.01400 [ 50]: 211112
...
As expected, the spin density is not sensitive for the two-antiferromagnetically coupled electrons
-------------------
SPIN-DENSITY MATRIX
-------------------
0 1 2 3 4 5
0 0.333333 0.000000 -0.000000 -0.000000 -0.000000 0.000000
1 0.000000 0.333333 -0.000000 0.000000 0.000000 0.000000
2 -0.000000 -0.000000 0.333333 -0.000000 0.000000 0.000000
3 -0.000000 0.000000 -0.000000 0.333333 -0.000000 0.000000
4 -0.000000 0.000000 0.000000 -0.000000 0.333333 0.000000
5 0.000000 0.000000 0.000000 0.000000 0.000000 0.333333
Trace of the spin density: 2.000000
The eff. unpaired density matrix by Head-Gordon on the other hand is able to reproduce the correct picture
here.
------------------------------------------------------------------------------
Head-Gordon eff. unpaired electron density
------------------------------------------------------------------------------
0 1 2 3 4 5
0 0.666667 0.000000 -0.000000 -0.000000 0.000000 -0.000000
1 0.000000 0.666667 0.000000 0.000000 0.000000 0.000000
2 -0.000000 0.000000 0.666667 -0.000000 0.000000 0.000000
3 -0.000000 0.000000 -0.000000 0.666667 -0.000000 0.000000
4 0.000000 0.000000 0.000000 -0.000000 0.666667 -0.000000
5 -0.000000 0.000000 0.000000 0.000000 -0.000000 0.666667
Number of eff. unpaired electrons ... 4.000
********************************
* MULLIKEN POPULATION ANALYSIS *
* (EFF. UED) *
********************************
------------------------------------------
MULLIKEN ATOMIC EFF. UED
------------------------------------------
0 O : 2.000000
1 O : 2.000000
Sum of atomic contributions: 4.0000000
---------------------------------------------------
MULLIKEN REDUCED ORBITAL EFF. UED
---------------------------------------------------
SPIN
0 O s : 0.000000 s : 0.000000
pz : 0.666667 p : 2.000000
px : 0.666667
py : 0.666667
dz2 : 0.000000 d : 0.000000
dxz : 0.000000
dyz : 0.000000
dx2y2 : 0.000000
dxy : 0.000000
1 O s : 0.000000 s : 0.000000
pz : 0.666667 p : 2.000000
px : 0.666667
py : 0.666667
dz2 : 0.000000 d : 0.000000
dxz : 0.000000
dyz : 0.000000
dx2y2 : 0.000000
dxy : 0.000000
The full printing will contain both eff. unpaired electron densities their Mulliken and Loewdin analysis.
Extensions for single-reference UHF/UKS wavefunction exist and might find their way into ORCA in the
future.
9.30.7 Local Spin Analysis
It is common practice in various areas of chemistry to think about the interaction of open-shell systems in
terms of local spin states. For example, in dimeric or oligomeric transition metal clusters, the exchange
coupling between open shell ions that exist locally in high-spin states is a much studied phenomenon.
Diradicals would be typical systems in organic chemistry that show this phenomenon. In quantum mechanics,
however, the total spin is not a local property, but instead a property of the system as a whole. The total
spin squared, S 2 , and its projection onto the z-axis, Sz , commute with the non-relativistic Hamiltonian and
hence, the eigenfunctions of the non-relativistic Hamiltonian can be classified according to good quantum
numbers S and M according to:
S2 SM = S(S + 1) SM

Sz SM = M SM

where SM is an exact eigenfunction of the non-relativistic Hamiltonian or an approximation to it that
conserves the total spin as a good quantum number. The total spin itself is given by the sum over the
individual electron spins as:
X
S= s(i)
i
And hence,
X
S2 = s(i)s(j)
i,j
is a two-electron property of the system. It is obviously not trivial to relate the chemically very meaningful
concept of local spin to a rigorous quantum mechanical treatment. While there are various proposals of how
to deal with this problem, we follow here a proposal of Clark and Davidson (Clark, A.E.; Davidson, E.R., J.
Chem. Phys. 2001, 115, 7382-7392). The following equations are implemented in the SCF and CASSCF
modules of Orca.
Clark and Davidson define fragment projection operators with the property:
PA PB = AB PA
and:
X
PA = 1
A
Then using this identity:
XX
S= s(i)PA(i)
i A
XX
S= s(i)PA (i)
A i
X
= SA (i)
A
they show that the local spin operators obey the standard relations for spin operators:
SA = SA
SA SA = i~SA
Hence
XX
S2 = SA SB
A B
But then importantly:
XX
SA SB = s(i)s(j)PA (i)PB (j)
i j
X XX
= 43 AB PA (i) + s(i)s(j){PA (i)PB (j) + PA (j)PB (i)}
A i j>i
With the first- and second-order density matrix:

Z
0
(x, x ) = N (x, x2 , ..., xN ) (x0 , x2 , ..., xN )dx2 ...dxN
!Z
N
(x1 , x0 1 ; x2 , x0 2 ) = (x1 , x2 , ..., xN ) (x0 1 , x0 2 , ..., xN )dx3 ...dxN
2
N
= 21 N (N 1)). Then:

(with 2
hSA SB i = 43 AB tr(PA ) + 2tr(PA (1)PB (2)s(1)s(2)(1, 1; 2, 2))
In terms of the number of electrons on site A and the expectation value of SzA

A 1
Sz = 2 tr( PA )

A
N = tr( + PA )
in terms of molecular orbitals:

A X
1
Sz = 2 pq hp|PA |qi
p,q
X +
NA =

pq hp|PA |qi
p,q
McWeeny and Kutzelnigg (McWeeny, R.; Kutzelnigg, W. Int. J. Quant. Chem. 1968, 11, 187-203) show that
for the expectation value of s(1)s(2), the relevant irreducible part of the two-body density can be expressed
in terms of the spinless density matrix of second order:
(0)
R0 (1, 10 ; 2, 20 ) = 13 (1, 10 ; 2, 20 ) 32 (2, 10 ; 1, 20 )
X
0 0 0 0
= 13 pq pq
rs p(1)q(1 )r(2)s(2 ) + 2rs p(2)q(1 )r(1)s(2 )
pqrs
X
0 0
= 31 (pq rq
rs + 2ps )p(1)q(1 )r(2)s(2 )
pqrs
3
with a normalization factor of 4 after spin integration. Hence using this:
(0)
hSA SB i = 34 AB tr(PA ) + 64 tr(PA (1)PB (2)R0 (1, 1; 2, 2))
And then performing the integral:

X
hSA SB i = 34 AB tr(PA ) 61
43 (pq rq A B
rs + 2ps )Ppq Prs
|{z} pqrs
1
2
This is the final and perhaps most compact equation. The projection operator can be defined in very many
different ways. The easiest is to L
owdin orthogonalize the basis set:
A X A 1/2
L = S
A
where L denotes the Lowdin basis. This means that molecular orbitals are expressed in the orthogonal basis
as:
cL = S+1/2 c
and the density as:
PL = S+1/2 PS+1/2
The fragment projector is defined as:
X
PA = |L i hL |
L A
Clark and Davidson suggest a slightly more elaborate projector in which first, the intra-fragment overlap is
eliminated. This happens with a matrix U that for two fragments takes form:
!
1/2
SA 0
U= 1/2
0 SB
where is the block of basis functions belonging to fragment A. Likewise:
!
+1/2
1 SA 0
U = +1/2
0 SB
Then the pre-overlap is:
= U SU
S
This contains the unit matrix in the intra-fragment blocks and non-zero elements elsewhere. This overlap
matrix is the finally orthogonalized to obtain the globally orthogonal L
owdin basis. We finally transform the
MO coefficients by the following transformation:
cL = S+1/2 U1 c
For the projectors, operating with the two MOs i and j gives:
X X
B A

A C L L
hi|PA |ji = L |L L |L ci c j
L A B C
L L
X X
= AB AC cL L
i c j
L A B C
L L
X
= cL L
i cj
L A
Herrmann et al. (Herrmann, C.; Reiher, M.; Hess, B.A. J. Chem. Phys. 2005, 122, 34102) give the correct
expression of the expectation values for a single spin-unrestricted determinant
( )
X X
hSA SB i = 3
4 AB PiiA + PiA
i
i i

X X X X X X
+ 1
4 PiiA Pjj
B
+ PiA B
i Pj
j PijA PijB PiAj PiBj PiA B
i Pjj PiiA PjBj
i i ij

ij j ij j i
j
X
PiAj PiBj
i
j
Which is used in the Orca implementation.
The use of the Local spin-implementation is very easy. All that is required is to divide the molecule
into fragments. The rest happens automatically. For example, let us consider two nitrogen atoms at the
dissociation limit. While the total spin state is S=0, the tow nitrogen atoms local exist in high-spin states
(S=3/2). Consider the following test job:
! HF def2-SVP UHF TightSCF PModel
%scf brokensym 3,3 end
* xyz 0 1
N(1) 0 0 0
N(2) 0 0 1094
*
and the output:
-------------------
LOCAL SPIN ANALYSIS (Loewdin* projector)
-------------------
(1) A.E. Clark; E.R. Davison J. Chem. Phys. (2001), 115(16), pp 7382-7392
(2) C. Herrmann, M. Reiher, B.A. Hess J. Chem. Phys. (2005) 122, art 034102-1
Number of fragments = 2
Number of basis functions = 28
Number of atoms = 2
... Fragment AO indices were mapped

... intra-fragment orthogonalization completed
... Global Loewdin orthogonalizer constructed
... Loewdin orthogonalized occupied orbitals constructed
<SA*SB> 1 2
----------------------------------
1 : 3.7568
2 : -2.2500 3.7568
<SzA> Seff(A)
--------------------------
1 : 1.5000 1.5017
2 : -1.5000 1.5017
thus perfectly corresponding to the expectations. The same can be done at the CASSCF level:
! HF def2-SVP UHF TightSCF PModel
%casscf nel 6 norb 6 nroots 1 end
* xyz 0 1
N(1) 0 0 0
N(2) 0 0 1094
*
With the result:
<SA*SB> 1 2
----------------------------------
1 : 3.7500
2 : -3.7500 3.7500
<SzA>* Seff(A)
--------------------------
1 : n.a. 1.5000
2 : n.a. 1.5000
* = for a singlet state all <SzA> values are zero by definition

Thus, cleanly confirming the expectations.
As a less trivial example, consider a typical Fe(III) antiferromatically coupled transition metal dimer. An
appropriate input may be:
! pbe def2-sv(p) def2/j tightscf kdiis soscf pmodel
%scf brokensym 5,5

end
* xyz -2 1
Fe(1) -1.93818 0.53739 -0.00010
Fe(2) 1.06735 0.47031 0.00029
S(3) -0.38935 2.59862 -0.00983
S(3) -0.48170 -1.59050 0.01091
S(1) 2.68798 0.43924 1.99710
S(1) 2.68692 0.42704 -1.99712
S(2) -3.55594 0.56407 -1.99889
S(2) -3.55850 0.58107 1.99646
H(1) 3.91984 0.39462 1.47608
H(1) 3.91940 0.39536 -1.47662
H(2) -4.78410 0.69179 -1.48280
H(2) -4.78991 0.49249 1.47983
*
Where one of the bridging sulfurs was assigned to each site respectively.
<SA*SB> 1 2
----------------------------------
1 : 7.7009
2 : -5.3721 7.7012
<SzA> Seff(A)
--------------------------
1 : 1.7579 2.3197
2 : -1.7579 2.3198
Nice shows the expected results with the local site spins being close to their ideal value 2.5 which would hold
for a high-spin Fe(III) ion.
9.30.8 UNO Orbital Printing
The analysis of UNOs can be controlled similarly. The flags together with their default values are shown
below:
%output
Print[ P_UNO_OccNum ] = 1; # Occupation numbers
Print[ P_UNO_AtPopMO_M ] = 0; # Mulliken atom pop.
# per UNO
Print[ P_UNO_OrbPopMO_M] = 0; # Mulliken orbital pop.
# per UNO
Print[ P_UNO_ReducedOrbPopMO_M] = 0;
# Mulliken reduced orbital
# pop. per UNO
Print[ P_UNO_AtPopMO_L ] = 0; # Loewdin atom pop.
# per UNO
Print[ P_UNO_OrbPopMO_L] = 0; # Loewdin orbital pop.
# per UNO
Print[ P_UNO_ReducedOrbPopMO_L] = 0;
# Loewdin reduced orbital
# pop. per UNO
end
9.31 Orbital and Density Plots
There are two types of graphics output possible in ORCA - two dimensional contour plots and three
dimensional surface plots. The quantities that can be plotted are the atomic orbitals, molecular orbitals,
natural orbitals, the total electron density or the total spin density. The graphics is controlled through the
block %plots.
9.31.1 Contour Plots
The contour plots are controlled via the following variables
%plots
#*** the vectors defining the cut plane
v1 0, 0, 0 # pointer to the origin
v2 1, 0, 0 # first direction
v3 0, 1, 0 # second direction
#*** alternative to defining vectors. Use atom coordinates
at1 0 # first atom defining v1
at2 2 # second atom defining v2
at3 4 # third atom defining v3
#*** resolution of the contour
dim1 45 # resolution in v2-direction
dim2 45 # resolution in v3-direction
#*** minimum and maximum values along v2 and v3
9.31 Orbital and Density Plots 731
min1 -7.0 # min value along v2 in bohr

max1 7.0 # min value along v2 in bohr
min2 -7.0 # min value along v3 in bohr
max2 7.0 # max value along v3 in bohr
#***
UseCol true # Use color in the plot (blue=positive,
# red=negative)
Skeleton true # Draw Skeleton of the molecule of those
# atoms that are in or close to the cut
# plane
Atoms true # Draw the atoms that are in the plane as
# circles
NCont 200 # Number of contour levels.
ICont 0 # Draw NCont equally space contours
1 # Start with 1/NCont and the double the
# value for each additional contour
#*** the format of the output file
Format Origin # straight ascii files
HPGL # plotter language files
#*** the quantities to plot
MO("MyOrbital-15xy.plt",15,0); # orbital to plot
v3= 0, 0, 1 # change cut plane
MO("MyOrbital-16xz.plt",16,0); # orbital to plot
ElDens("MyElDens.plt"); # Electron density
SpinDens("MySpinDens"); # Spin density
end
The input was:
v1 = 0, 0, 0;
v2 = 1, 0, 0;
v3 = 0, 1, 0;
min1= -8; max1= 8;
min2= -8; max2= 8;
dim1= 50; dim2=50;
Format = HPGL;
NCont = 200;
Icont = 1;
Skeleton= true;
Figure 9.47: Contour plot of the lowest unoc- Atoms = true;
cupied spin down orbital of the MO("Test-DFT-H2CO+-MO7xy.plt",7,1);
H2 CO+ cation radical in the x, y
plane.
NOTE:
The command MO("MyOrbital-15xy.plt",15,0); is to be interpreted as follows: MO means that a

MO is to be plotted. MyOrbital-15xy.plt is the file to be created. 15 is the number of the MO to
be drawn (remember: counting starts at orbital 0!) and 0 is the operator the orbital belongs to. For
a RHF (or RKS) calculation there is only one operator which has number 0. For a UHF (or UKS)
calculation there are two operators - the spin-up orbitals belong to operator 0 and the spin-down
orbitals belong to operator 1. For ROHF calculations there may be many operators but at the end all
orbitals will be collected in one set of vectors. Thus the operator is always =0 in ROHF.
The ELDENS (plot of the total electron density) and SPINDENS (plot of the total spin density) commands
work analogous to the MO with the obvious difference that there is no MO or operator to be defined.
The UNO option plots natural orbitals of the UHF wavefunction (if they are available). No operator can
be given for this command because there is only one set of UHF-NOs.
If the program cannot find the plot module (Bad command or filename) try to use ProgPlot="orca plot.exe"
in the %method block or point to the explicit path.
The defining vectors v2 and v3 are required to be orthonormal. The program will use a Schmidt
orthonormalization of v3 with respect to v2 to ensure orthonormality. If you do not like this make
sure that the input vectors are already orthogonal.
at1, at2 and at3 can be used instead of v1, v2 and v3. In this case say v1 is taken as the coordinates
of atom at1. Mixed definitions where say v2 is explicitly given and say v3 is defined through at3 are
possible. A value of -1 for at1, at2 and at3 signals that at1, at2 and at3 are not to be used. This
type of definition may sometimes be more convenient.
Variables can be assigned several times. The actual value a variable has is stored together with the
command to generate a plot (MO, ELDENS or SPINDENS). Thus after each plot command the format or
orientation of the plot can be changed for the next one.
The Origin format produces a straightforward ASCII file with x, y and z values that can be read
into your favorite contour plot program or you could write a small program that reads such files and
converts them to whatever format is more appropriate for you.
I usually use Word for Windows to open the HPGL files which appears to work fine. Double clicking on
the graphics will allow modification of linewidth etc. For some reason that is not clear to me some
graphics programs do not like the HPGL code that is produced by ORCA. If you are an HPGL expert
and you have a suggestion - let me know.
9.31.2 Surface Plots
9.31.2.1 General Points
Surface plots can, for example, be created through an interface to Leif Laaksonens gOpenMol program.
This program can be obtained free of charge over the internet (https://research.csc.fi/-/gopenmol). It
runs on a wide variety of platforms, is easy to use, produces high quality graphics and is easy to interface22 -
thank you Leif for making this program available!
The relevant [PLOTS] section looks like this:
%output
XYZFile true
end
%plots
dim1 45 # resolution in x-direction
dim2 45 # resolution in y-direction
dim3 45 # resolution in z-direction
min1 -7.0 # x-min value in bohr
max1 7.0 # x-min value in bohr
min2 -7.0 # y-min value in bohr
max2 7.0 # y-max value in bohr
min3 -7.0 # z-min value in bohr
max3 7.0 # z-max value in bohr
Format gOpenMol_bin # binary *.plt file
gOpenMol_ascii # ascii *.plt file
Gaussian_Cube # Gaussian-cube format
# (an ASCII file)
MO("MyOrbital-15.plt",15,0); # orbital to plot
MO("MyOrbital-16.plt",16,0); # orbital to plot
UNO("MyUNO-48.plt",48); # UHF-NO to plot
ElDens("MyElDens.plt"); # Electron density
SpinDens("MySpinDens.plt"); # Spin density
end
NOTE:
it is admittedly inconvenient to manually input the dimension of the cube that is used for plotting. If
you do nothing such that min1 = max1 = min2 = max2 = min3 = max3=0 then the program will
try to be smart and figure out a good cube size by itself. It will look at the minimum and maximum
values of the coordinates and then add 7 bohrs to each dimension in the hope to properly catch all
wavefunction tails.
Sometimes you will want to produce orbital plots after you looked at the output file and decided which
orbitals you are interested in. In this case you can also run the orca plot program in a crude interactive
form by invoking it as:
orca_plot MyGBWFile.gbw -i
22
There were some reports of problems with the program on Windows platforms. Apparently it is better to choose
the display settings as true color 32 bit rather than high 16 bit. Thanks to Thomas Brunold!
Figure 9.48: The total electron density (shown as a mesh) and the spin density (shown with solid
contours) of the H2 CO+ cation radical as calculated by the RI-BP/VDZP method.
Note the small negative spin density at the carbon atom. The spin density was cal-
culated at 120x120x120 resolution which takes much longer than the DFT calculation
itself.
This will only provide you with a small subset of the capabilities of this program but may already be enough
to produce the plots you want to look at. Note that for the name of the GBW-file you may as well input
files that result from natural orbitals (normally *.uno), corresponding orbitals (normally *.uco) or localized
orbitals (normally *.loc).
Figure 9.49: The orbital of H2 CO as calculated by the RI-BP/VDZP method.
9.31.2.2 FOD plots
The fractional occupation number weighted electron density (F OD , see 9.5.8.1) can be plotted in 3D for a pre-
defined contour surface value which, after extensive testing, was set to the default value of = 0.005 e/Bohr3 .
In order to allow comparison of various systems this value should be kept fix (in critical cases, one may
also check the FOD plot with a a smaller value of = 0.002 e/Bohr3 for comparison). The FOD is strictly
positive in all space and resembles orbital densities (e.g., -shape in large polyenes) or the total charge density
for an ideal metal with complete orbital degeneracy in trivial cases. FOD plots represent a cost-effective
and robust way to identify the hot (strongly correlated) electrons in a molecule and to choose appropriate
approximate QC methods for a subsequent computational study of the systems in question. Based on our
experience, the following rules of thumb can be derived:
a) no significant F OD : use (double)-hybrid functionals or (DLPNO-)CCSD(T) (single-reference electronic

structure)
b) significant but rather localized F OD : use semi-local GGA functionals (or hybrid functional with low
Fock-exchange, avoid HF or MP2; slight multi-reference character)
c) significant and delocalized F OD : use multi-reference methods (or finite temperature DFT; strong
multi-reference character)
Basically, F OD can be plotted analogously to an electron density calculated with ORCA using Basename.scfp fod
instead of Basename.scfp. Note that producing *.cube files with orca plot (see 9.32.7) may take a consid-
erable amount of time for larger molecules, particularly if high quality plots for publication purposes (i.e.,
120x120x120 resolution) are wanted. An example FOD plot (singlet ground sate of p-benzyne, see 9.5.8.1 for
the corresponding ORCA input) is shown in Fig. 9.50. It has been produced with the UCSF CHIMERA pro-
gram (this program can be obtained free of charge over the internet: https://www.cgl.ucsf.edu/chimera/)
using the *.cube file generated with orca plot:
orca_plot pbenzyne.gbw -i
user input:
1 (type of plot)
2 (electron density)
n (default name: no)
pbenzyne.scfp fod (name of the FOD file)
4 (number of grid intervals)
120 (NGrid)
5 (output file format)
7 (cube)
10 (generate plot)
11 (exit)
It is also possible to generate *.cube files from F OD (analogously to electron density plots) with other pro-
grams that can read ORCA BaseName.gbw and electron density files by simply using the Basename.scfp fod
file instead of the Basename.scfp file.
The significant and rather delocalized FOD for p-benzyne (1 Ag ) indicates that multi-reference methods would
be needed for reliable computational study of this molecule (category c)). More examples of FOD plots
generated with the same setup and programs can be found in the original publication and corresponding
supplementary information. [307]
Figure 9.50: FOD plot at = 0.005 e/Bohr3 (TPSS/def2-TZVP (T = 5000 K) level) for the 1 Ag
ground state of p-benzyne (FOD depicted in yellow).
9.31.2.3 Interface to gOpenMol
Here is a short summary of how to produce these plots with gOpenMol:
First of all the molecular geometry must be save by choosing XYZFile=true in the [OUTPUT] block.
This will produce a straightforward ascii file containing the molecular geometry (or simply ! XYZFile).
After having produced the plot files start gOpenMol and choose File-Import-Coords. In the dialog
choose the XYZ format and select the file. Then press apply and dismiss. The molecule should now
be displayed in the graphics window.
You can change the appearance by choosing View-Atom type.
The color of the background can be changed with Colour-Background .
After having done all this choose Plot-Contour and select the Browse button to select the appropriate
file. Then press Import File to read it in. NOTE: you can only directly read files that were produced
in gOpenMol bin format. If you have chosen gOpenMol ascii you must first use the gOpenMol file
conversion utility under Run-Pltfile (conversion) to produce the binary plt file.
After having read the plt file choose the appropriate isocontour value (one for the positive and one for
the negative lobes of an orbital) and select suitable colors via Colour(n) to the right of the isocontour
value. The Details button allows you to choose between solid and mesh representation and other
things.
Once the plot looks the way you like, use File-Hardcopy to produce a publication quality postscript
or bitmap picture that can be imported into any word processing or graphics software.
9.31.2.4 Interface to Molekel
The Molekel program (http://ugovaretto.github.io/molekel/) is another beautiful and easy-to-use

graphics tool that is recommended in combination with ORCA. You may even find it a little easier to use
than gOpenMol but this may be a matter of personal taste. In order to produce plots with Molekel follow
the following procedure:
9.32 Utility Programs 737
Produce Gaussian-Cube files (and optionally also an XYZ file) with ORCA as described above.
Start Molekel and use the right mouse button to obtain the Load menu.
Choose the format xyz to load your coordinates
Use the right mouse button again to select the Surface menu
Choose the format Gaussian Cube and click Load Surface
Click on Both Signs if you visualize an orbital or spin density
Select a suitable contour value in the Cutoff box.
Click on Create Surface. Thats it!
In the Color menu (also available via the right mouse button) you can adjust the colors and in the
main menu the display options for your molecule. Default settings are in a startup file that you can
modify to suit your taste. More details are in the Molekel manual check it out; it can do many other
useful things for you too!
9.32 Utility Programs
9.32.1 orca mapspc
This utility program is used to turn calculated spectra into a format that can be plotted with standard
graphics programs. The usage is simple (for output examples see for example sections 9.18.2, 8.13.3.1, 8.13.3.2
and 9.19.4):
orca_mapspc file spectrum options
file = name of an ORCA output file

name of an ORCA hessian file (for IR and Raman)
spectrum= abs - Absorption spectra

cd - CD spectra
ir - IR spectra
raman - Raman spectra
options -x0value: Start of the x-axis for the plot

-x1value: End of the x-axis for the plot
-wvalue : Full-width at half-maximum height in
cm**-1 for each transition
-nvalue : Number of points to be used
The exact abilities of orca mapspc can be seen by simply executing the command in a terminal
orca_mapspc
Then one gets:
----------------------------------------------------------------------------------
usage: orca_mapspc Output-file ABS, ABSV, ABSQ, ABSOI, CD, IR, RAMAN, NRVS, VDOS,
MCD, SOCABS, SOCABSQ,SOCABSOI, XES, XESV, XESQ, XESOI,
XAS, XASV, XASQ, XASOI, XESSOC,XASSOC, RIXS, RIXSSOC -options
----------------------------------------------------------------------------------
----General options ----
-o output file
-cm use cm**-1 (default)
-eV use eV (default cm**-1)
-g use Gaussian lineshape default
-l use Lorentz lineshape (only for Absorption and Emission like spectra)
-v use Voigt lineshape (only for Absorption and Emission like spectra)
-x0 initial point of spectrum
-x1 final point of spectrum
-w line width for Gaussian/Lorenzian linewidth
-q line width for the Gaussian part of Voigt linewidth
-kw coeffitient for the line width calculated as kw*sqrt(energy)
-n number of points
----The following additional options are for RIXS and RIXSSOC calculations----
-x2 initial point of the spectrum along y axis
-x3 final point of the spectrum along yaxis
-g line width for Gaussian/Lorenzian linewidth along y axis
-m number of points for the emission spectrum
-eaxis plot option for the emission axis: (1) for Energy transfer
(2) for emission spectrum
-uex number of user defined cuts at constant Excitation Energy axis
-udw number of user defined cuts at constant Emission/ Energy Transfer axis
-dx number for shifting the spectra along the Excitation /Emission Energy axis
-kg coeffitient for the line width calculated as kg*sqrt(energy)
----Using external files----
paras.inp: a list of energy ranges with desired broadening parameters
for x axis: E_start E_stop Width
for y axis: 0 0 0 E_start E_stop Width
for xy axis: E_start1 E_stop1 Width1 E_start2 E_stop2 Width2
udex.inp: list of energies for taking cuts
at constant Excitation Energy axis (RIXS/RIXSOC)
udem.inp: list of energies for taking cuts
at constant Emission/ Energy Transfer axis (RIXS/RIXSOC)
gfsp.inp: list of ground-final state pairs to generate
individual state pair RIXS planes
and respective analysis planes (ROCIS RIXS/RIXSOC)

---------------------------------------------------------------------------------
NOTE:
The input to this program can either be a normal output file from an ORCA calculation or a ORCA
.hess file if ir or Raman spectra are desired
Unless it is specified otherwise the default lineshape is always assumed to be a gaussian
There will be two output files:
Input-file.spc.dat (spc=abs-like or cd or ir or raman): This file contains the data to be

plotted
Input-file.spc.stk: This file contains the individual transitions (wavenumber and intensity)
The absorption plot has five columns: The first is the wavenumber in reciprocal centimeters, the second
the total intensity and the third to fifth are the individual polarizations (i.e. assuming that the electric
vector of the incoming beam is parallel to either the input x-, or y- or z-axis respectively). The last
three columns are useful for interpreting polarized single crystal spectra.
Generation of multiple spectra. When more than one spectra of the same kind are available the
program will try to plot them. For example in the case of a CASSCF calculation with the NEVPT2
flag on, there will be two Absorption spectra (CASSCF and NEVPT2) that can be ploted
For example:
orca_mapspc My-CASSCF/NEVPT2-Output.out SOCABS -x07000 -x18000 -eV -n10000 -w2.0 -l
Mode is SOCABS
Entering SOC-ABS reading
Using eV units
Using Lorentzian shape
Multiple SOCABS (2) spectra detected ...
----------------------------------
Plotting SOCABS Spectrum 0
----------------------------------
Start energy [eV] ... 7000.00
Stop energy [eV] ... 8000.00
Peak FWHM [eV] ... 2.00
----------------------------------
Plotting SOCABS Spectrum 1

----------------------------------
Start energy [eV] ... 7000.00
Stop energy [eV] ... 8000.00
Peak FWHM [eV] ... 2.00
This will generate two kind of spectra one for the CASSCF and one for the NEVPT2 calculation
CASSCF:
My-CASSCF/NEVPT2-Output.out.0.socabs.dat
My-CASSCF/NEVPT2-Output.out.0.socabs.stk
NEVPT2:
My-CASSCF/NEVPT2-Output.out.1.socabs.dat
My-CASSCF/NEVPT2-Output.out.1.socabs.stk
Other Absorption or CD spectra can also be generated in the same way.
9.32.2 orca chelpg
This program calculates CHELPG atomic charges according to Breneman and Wiberg [420]. The atomic
charges are fitted to reproduce the electrostatic potential on a regular grid around the molecule, while
constraining the sum of all atomic charges to the molecules total charge.
The program works with default values in the following way:
orca_chelpg MyJob.gbw
The program uses three adjustable parameters, which can also be set in a separate chelpg input block
%chelpg
GRID 0.3 # Spacing of the regular grid in Ang
RMAX 2.8 # Maximum distance of all atoms to any
# gridpoint in Ang
VDWRADII COSMO # VDW Radii of Atoms, COSMO default
BW # Breneman, Wiberg radii
end
In this case the ORCA automatically calculates the CHELPG charges at the end of the calculation. Automatic
calculation of CHELPG charges using the default values can also be achieved by specifying
! CHELPG
In the simple input section. By default the program uses the COSMO VDW radii for the exclusion of
gridpoints near the nuclei, as these are defined for all atoms. The BW radii are similar, but only defined for
very few atom types.
9.32.3 orca pltvib
This program is used in conjunction with gOpenMol (or xmol) to produce animations or plots of vibrational
modes following a frequency run. The usage is again simple and described in section 8.13.3.5 together with a
short description of how to produce these plots in gOpenMol.
The program produces 20 frames of animation, where first and last frame correspond to the TS, all others
calculated as sin(2f rame/20 1) displacement.
9.32.4 orca vib
This is a small standalone program to perform vibrational analysis. The idea is that the user has some
control over things like the atomic masses that enter the prediction of vibrational frequencies but are
independent of the electronic structure calculation as such.
The program takes a .hess file as input and produces essentially the same output as follows the frequency
calculation. The point is that the .hess is a user-editable textfile that can be manually changed to achieve
isotope shift predictions and the like. The usage together with an example is described in section 8.13.3.6. If
you pipe the output from the screen into a textfile you should also be able to use orca mapspc to plot the
modified IR, Raman and NRVS spectra.
9.32.5 orca loc
Localization is a widely used technique nowdays. By defining different functionals, various localization
methods are established. The most favorable localization methods are developed by Foster-Boys and Pipek-
Mezey. In ORCA there are four different localization methods available, the Pipek-Mezey method (PM),
the Foster-Boys method (FB), the intrinsic atomic orbitals (IAO) based PM method and the IAO based FB
method. (There is another extra FB algorithm, which could be faster to localize virtual MOs of very large
systems.)
The intrinsic atomic orbitals and intrinsic bond orbitals (IAOIBO) localization method is developed by
Gerald Knizia, see Ref. [421]. In IAOIBO method, the occupied MOs are projected to a minimal basis set to
get the IAOs, firstly. In ORCA different from original IAOIBO method, the converged SCF MO of atoms
are used instead of Huzinaga MINI or STO-3G. However, the IAO charges computed by our method are
quite similar to original IAO. Then, Pipek-Mezey functional is employed to localize these IAOs to IBOs.
Finally,IBOs will be backtransformed to their original basis set. The IAO partial charges of canonical MOs for
each atom is also printed out before the IAOIBO localization. But make sure you are included all occupied
MOs in the IAOIBO localization. Otherwise, the IAO charges are meaningless. We further improved the
original IAOIBO method by using the FB functional instead of PM functional. The computational time
of the new method named IAOBOYS should be faster than the standard FB method for large systems.
However, the IAO based method can only be used for the localization of occupied MOs.
There are two ways to do the MO localization in ORCA . The simpler way is to request the localization at
the end of any ORCA calculation input file. Details are set in the %loc block.
%loc
LocMet PM # Localization method e.g. PIPEK-MEZEY
FB # FOSTER-BOYS
IAOIBO # IAOIBO
IAOBOYS # IAOBOYS
NEWBOYS # FOSTER-BOYS
Tol 1e-6 # absolute convergence tolerance for the localization sum
# default value is 1e-6
Random 0 # Always take the same seed for start for localization
# (For testing/debug purpose,optional)
1 # Take a random seed for start of localization (default)
PrintLevel 2 # Amount of printing
MaxIter 64 # Max number of iterations
T_Bond 0.85 # Thresh that classifies orbitals in bond-like at the printing
T_Strong 0.95 # Thresh that classifies orbitals into strongly-localized at
# the printing
OCC true # Localize the occupied space
T_CORE -99.9 # The Energy window for the first OCC MO to be localized (in a.u.)
# Here, we localize all occupied MOs including core orbitals.
VIRT true # Localize the virtual space
end
The localized MOs are obtained iteratively. Convergence is achieved when the localization functional value
is self-consistent (contraled by Tol). Setting the flags OCC/VIRT to true will request a localization of
the subspace. If both flags are set, two consecutive localizations are performed. The localized orbitals are
stored in the form of a standard GBW file named .loc. Keep in mind that the localization of the occupied
orbitals might change the total energy depending on what type of calculation you want to perform thereafter.
For RHF and UHF there shouldnt be any problems, but for CASSCF the keyword OCC is not sufficient.
CASSCF is not invariant to rotation of all the occupied orbitals.
The other way to do the localization is calling the orca loc program directly from shell, which is more
general. The orca loc program requires an input of its own. The input is a textfile containing the necessary
parameters. If no input is specified, orca loc returns a help-file with a description of the necessary input-
parameters. You need to specify in/output gbw-files, along with orbital ranges and the localization method
to be used. A source of confusion is the operator line op (alpha = 0 or beta = 1). For RHF(ROHF) and
CASSCF, this should be set to zero. The input file usually looks like,
Myjob.gbw # input orbitals

Myjob.loc.gbw # output orbitals
0 # operator:0 for alpha, 1 for beta
10 # orbital window: first orbital to be localized e.g. first active
15 # orbital window: last orbital to be localized e.g. last active
200 # maximum number of iterations
1e-6 # convergence tolerance of the localization functional value
0.0 # relative convergence tolerance of the localization functional value
0.95 # printing thresh to call an orbital strongly localized
0.85 # printing thresh to call an orbital bond-like
4 # localization method:
# 1=PIPEK-MEZEY,2=FOSTER-BOYS,3=IAO-IBO,4=IAO-BOYS, 5=NEW-BOYS
1 # use Cholesky Decomposition (0=false, 1=true, default is true,optional)
0 # Randomize seed for localization(optional)
If the input file is called myloc.inp, running orca loc myloc.inp will produce the Myjob.loc.gbw file containing
the localized orbitals. Please make sure the Myjob.gbw is in the same directory as myloc.inp.
9.32.6 orca blockf
This utility program allows the canonicalization of orbitals (.gbw file) for arbitrary subspaces. With
canonicalization we refer to the block diagonalization of the Fock matrix. Note that the necessary Fock
matrix must be generated and be available on disk prior calling orca blockf. The program is described in
section 9.23.2.2, where the Local ZFS decomposition is discussed.
9.32.7 orca plot
The use of this program is described more fully in section 9.31. It is used to create three dimensional graphics
data for visualization. It is also possible to run this program interactively:
orca_plot gbwfile -i
You will then get a simple, self-explaining menu that will allow you to generate a variety of files (such as
.plt and .cube) directly from the .gbw files without restarting or running a new job.
It is possible to use the module to create difference densities between the ground and excited states from
CIS or TD-DFT calculations. This is implemented as an extra interactive menu point that is (hopefully)
self-explanatory.
9.32.8 orca 2mkl: Old Molekel as well as Molden inputs
This little utility program can be used to convert gbw files into mkl files which are of ASCII format. This is
useful since molekel can read these files and use them for plotting and the like. The contents of the mkl file
is roughly the same as the gbw file (except for the internal flags of ORCA) but this is an ASCII file which
can also be read for example by your own programs. It would therefore be a good point for developing an
interface. It is likely that this functionality will be further expanded in the future.
orca_2mkl BaseName
(will produce BaseName.mkl from BaseName.gbw)
orca_2mkl BaseName -molden
(writes a file in molden format)
orca_2mkl BaseName -mkl
(writes a file in MKL format)
We have recently also added the capability to convert any gbw type file into MKL or Molden format. Thus,
you can use this device to vizualize QRO or UNO or UCO orbitals or any type of natural orbitals:
orca_2mkl gbw_type_file.extension mkl_file.extension -mkl -anyorbs

# or
orca_2mkl gbw_type_file.extension molden_file.extension -molden -anyorbs
You also have the opportunity to run orca 2mkl backwards in order to produce gbw type files. You can use
this device in order to import orbitals from other sources into ORCA. This is not a frequently used option
and it has limited capabilities. Hence, it is documented here only in a cursory way in order for you to be
able to experiment. Note that the CASSCF tutorial, that supplements the manual, shows how to edit the
molecular orbitals using orca 2mkl.
orca_2mkl BaseName -gbw

(will produce BaseName.gbw out of BaseName.mkl)
9.32.9 orca 2aim
This utility program reads a .gbw file and creates a .wfn file that can be used for topological analysis of the
electron density by other programs. Currently this works only for closed-shell wave functions. The usage is
very simple just type AIM in the simple input line of your input file, or use
orca_2aim BaseName
(will produce BaseName.wfn from BaseName.gbw)
9.32.10 orca vpot
This program calculates the electrostatic potential at a given set of user defined points. It takes four
arguments:
orca_vpot MyJob.gbw MyJob.scfp.tmp MyJob.vpot.xyz MyJob.vpot.out
First: The gbw file containing the correct geometry and basis set
Second : The desired density matrix in this basis (perhaps use the KeepDens keyword)
Third : an ASCII file with the target positions, e.g.
6 (number_of_points)
5.0 0.0 0.0 (XYZ coordinates)
-5.0 0.0 0.0
0.0 5.0 0.0
0.0-5.0 0.0
0.0 0.0 5.0
0.0 0.0 -5.0
Fourth: The target file which will then contain the electrostatic potential, e.g.
6 (number of points)
VX1 VY1 VZ1 (potential for first point)
VX2 VY2 VZ2 (potential for second point)
etc.
It should be straightforward for you to read this file and use the potential for whatever purpose.
9.32.11 orca euler
This utility program is used to calculate the relative orientation between calculated hyperfine coupling
(HFC)/nuclear quadrupole coupling (NQC) tensors and a reference tensor (the calculated molecular g-/D-
tensor). The orca euler program is run by default in an ORCA job after the calculation of HFCs or NQCs,
if g- or D-tensor are also calculated in the same job. The utility program can also be run as a stand-alone
program. In this case the .prop file of a previous NQC/HFC- and D- or g-tensor calculation must be
available.
The orientation between the tensors is calculated in terms of a 3x3 rotational matrix R. This is parametrized
by the three so-called Euler angles , and . These angles define the relative orientation between two
tensors A and B by three successively applied rotations around different axes in order to align A with B. In
the commonly used z-y-z convention these three rotations are:
Rotate Axyz counterclockwise around its z axis by to give Ax0 y0 z0 .
Rotate Ax0 y0 z0 counterclockwise around its y 0 axis by to give Ax00 y00 z00 .
Rotate Ax00 y00 z00 counterclockwise around its z 00 axis by to align with B.
orca_euler prop-file options
file = name of an ORCA .prop file
options
-refg/-refD: Reference tensor (g-tensor or D-tensor, default is -refg)
-conv zyz/-conv zxz: Euler rotation convention (default is zyz)
-order: Ordering of the reference tensor (x, y, z) with respect to
ORCA output (min, mid, max)
-plotA: plot the HFC-tensors
-plotQ: plot the NQC-tensors
-detail: print detailed information
NOTE:
By default the D-tensor is used as reference tensor only if S > 12 and if |D|>0.3 cm1 ; in all other cases
the g-tensor is used as reference tensor. The user can manually select the reference tensor if the
information is available in the prop-file by using refg or -refD.
By default the Euler rotation in the z-y-z convention is used. The z-x-z convention can be selected
manually by using the option conv zxz.
By default the axes of the g- or D-tensor are assigned depending on their magnitude. gmin gx ,
gmid gy , gmax gz (similarly for D). This ordering can be modified manually when running the
standalone program as shown in the following examples:
-order 3 2 1: min z
mid y
max x
-order 1 -2 3: min x
mid y (flipped in the orientation)
max z
The nuclear hyperfine and quadrupole coupling tensors can be plotted (in the xyz-file format) by the
orca euler program using plotA or plotQ. The HFC tensor for atom 3 (counting starts at zero) is
e.g. stored in the file prop-file.3.A.xyz, the respective NQC tensor is stored in prop-file.3.Q.xyz.
In these xyz files the position of four atoms (He, Ne, Ar, Kr) is given. The x-, y- and z-direction of the
tensor are in the direction of the vectors between He-Ne, He-Ar and He-Kr.
The actual definition of the used rotation matrix and more information on the relative orientation can
be printed by using the option detail.
9.32.12 orca exportbasis
A small utility program to print out the basis sets used by ORCA. Its usage requires at least the name of the
basis set, as specified in the simple input line of ORCA. Additional parameters like an ECP basis set, a list
of specific atoms or the name of an ouput file are accepted. The output is stored in ASCII format, it can be
inspected and modified. The user can choose to print the basis sets in either ORCA format, which then can
be copied into the input file, or in GAMESS-US format, which can be read via the %basis block as externally
specified basis. NOTE: Basis set names containing special characters may need a pair of enclosing or to
be recognized.
USAGE: orca_exportbasis keywords options
-b, --basis : name of basis set

def2-svp
def2-tzvp(-f) - string to be passed with literals
EXAMPLE: orca_exportbasis -b svp
Additional Options:
-e, --ecp : ecp basis

sdd
default - ECP-part of basis (if present)
-f, --format : output format

ORCA - to be read via %basis NewGTO
GAMESS-US - to be read as %basis GTOName mybasis.bas
default - ORCA
-a, --atoms : list of elements
Cu - single element
Ga Ge As Se - list of elements separated by blanks
default - whole periodic table is printed
-o, --outfile: name of outputfile

mybasis.bas
default - derived name
EXAMPLE: orca_exportbasis -b svp -e sdd -a Ag -f GAMESS-US -o mybasis.bas
The output stored in GAMESS-US format can be used in the %basis block of the next ORCA calculation.
%basis
GTOName = "mybasis.bas";
GTOAuxName = "myauxbasis.bas";
end
9.32.13 orca eca
This utility program makes use of the calculated exchange coupling constants to compute relative energies of
all possible spin states through diagonalization of spin Hamiltonian. The absolute and relative energies of the
spin states are printed in the *.en and *.en0 files respectively. The on-site spin expectation values are also
printed in a *.sp file.The following example calculates the spin ladder for a system with exchange coupling
constant of -152.48 cm**-1 between Mn(III) and Mn(IV).
%sim
ms_bs 0.5 # Arbitrary spin state
end
# specification of spin centers

$spins 2
1 2.0 # Spin on first manganese
2 1.5 # Spin on second manganese
# Exchange coupling constant (H = -2J S1 S2)

$ecc 1
1 -152.48
$aiso_bs 2 # A false segment just to print the *.sp file

1 0.00
2 0.00
749
10 Some Tips and Tricks
10.1 Input
For calculations on open-shell systems we recommend to use the keywords !UNO !UCO in the input line. This
will generate quasi-restricted molecular orbitals QRO, unrestricted natural spin-orbitals UNSO, unrestricted
natural orbitals UNO and unrestricted corresponding orbitals UCO. Moreover, it will print the UCO overlaps
in the output, which can provide very clear information about the spin-coupling in the system. Below an
example of the input and section of the output is provided.
!B3LYP def2-SVP def2/J RIJCOSX UNO UCO TightSCF
The UCO overlap section in the output will look like:
***UHF Corresponding Orbitals were saved in MyJob.uco***
----------------------
Orbital Overlap(*)
----------------------
.
.
.
96: 0.99968
97: 0.99955
98: 0.99947
99: 0.99910
100: 0.99873
101: 0.99563
102: 0.74329
103: 0.00000
The overlap corresponds to a value usually less than 0.85 denotes a spin-coupled pair. Whereas, values close
to 1.00 and 0.00 refers to the doubly occupied and singly occupied orbitals respectively.
10.2 Cost versus Accuracy
A difficult but important subject in electronic structure theory is to balance the price/accuracy ratio of the
calculations. This ratio is governed by: (a) the method used, (b) the basis set used and (c) the cutoffs and
tolerances used. There are certainly differing opinions among scientists and I merely quote a few general,
subjective points:
750 10 Some Tips and Tricks
Calculations with minimal basis sets are always unreliable and are only good for explorations. This
is also true for small split-valence basis sets like 3-21G, 3-21GSP and perhaps also 4-22GSP. These
basis sets are significantly more reliable than their minimal basis counterparts but are not capable of
delivering quantitatively reliable results. They may, however, be the only choice if very large molecules
are targeted.
In our own research we almost exclusively use the basis sets of the Karlsruhe group for non-relativistic
calculations. They have been updated to the def2 set that is more consistent than the older basis
sets.
Def2-SV(P) is the smallest and computationally efficient split-valence basis set and is largely identical
to the old SV(P), except for the transition metals which have more consistent polarization sets.
Def2-TZVP is different from the old TZVP. It has been realized that if one invests into an accurate
triple-zeta description of the valence region it makes limited sense to only employ a single polarization
function. The accuracy is then limited by the polarization set and is not much better than what one
gets from SV(P). Hence, def2-TZVP contains a single p-set for hydrogens but is otherwise very similar
to the old TZVPP basis set, e.g. it contains 2d1f polarization for main group elements and much more
extensive polarization sets for transition metals. The highest polarization function (f for main group)
does add substantially to the computational effort. Hence, we often use def2-TZVP without the f
polarization function. In order to do that one can use the keyword def2-TZVP(-f). Together with RI
or RIJCOSX this is still computationally economic enough for most studies.
Def2-TZVPP is a fully consistent triple-zeta basis set that provides excellent accuracy for SCF
calculations (HF and DFT) and is still pretty good for correlated calculations. It is a good basis set to
provide final single point energies.
Def2-QZVPP is a high accuracy basis set for all kinds of calculations. It provides SCF energies near
the basis set limit and correlation energies that are also excellent. It is computationally expensive
but with RI and RIJCOSX in conjunction with parallelization it can often still be applied for final
single-point energy calculations. In conjunction with such large basis sets one should also increase the
accuracy of the integration grids in DFT and RIJCOSX it would be a shame to limit the accuracy
of otherwise very accurate calculations by numerical noise due to the grid.
Correlation consistent basis sets provide good correlation energies but poor to very poor SCF energies.
For the same size, the ano-PVDZ basis sets are much more accurate but are also computationally more
expensive. Except for systematic basis set extrapolation we see little reason to use the cc bases.
Pople basis sets are somewhat old fashioned and also much less consistent across the periodic table
than the basis from the Karlsruhe group. Hence, we generally prefer the latter.
For scalar relativistic calculations (ZORA and DKH) we strongly recommend to use the SARC bases
in conjunction with the ZORA or DKH recontractions of the Karlsruhe bases. They are also flexible
enough in the core region for general purpose and spectroscopic applications.
Effective core potentials lead to some savings (but not necessarily spectacular ones) compared to
all-electron relativistic calculations. For accurate results, small core ECPs should be used. They are
generally available for the def2 Karlsruhe type basis sets for elements past Krypton. In general we
prefer StuttgartDresden ECPs over LANL ones. For the first transition row, the choices are more
meager. Here Karlsruhe basis sets do not exist in conjunction with ECPs and you are bound to either
10.2 Cost versus Accuracy 751
SDD or LANL of which we recommend the former. Geometries and energies are usually good from
ECPs, but for property calculations we strongly recommend to switch to all electron scalar relativistic
calculations using ZORA (magnetic properties) or DKH (electric properties).
You can take advantage of a built-in basis set (Print[P Basis]=2) and then modify it by uncontracting
primitives, adding steeper functions etc. (fully uncontracted bases are generated via uncontract in
%basis) Alternatively some basis sets exist that are of at least double-zeta quality in the core region
including the DZP and Dunning basis sets. For higher accuracy you may want to consider the aug-
series of basis sets.
Likewise, if you are doing calculations on anions in the gas phase it is advisable to include diffuse
functions in the basis set. Having these diffuse functions, however, makes things much more difficult as
the locality of the basis set is significantly reduced. If these functions are included it is advisable to
choose a small value for Thresh (1012 or lower).
In case of charged molecules, implicit solvent models are usefull in providing more realisitic charge
distributions and energetics. The use of CPCM model is recommended.
The integration grids used in DFT should be viewed together with the basis set. If large basis set
calculations are converged to high accuracy it is advisable to also use large DFT integration grids (like
Grid=5 or even Grid=6). For unlimited accuracy (i.e. benchmark calculations) it is probably best
to use product grids (Grid=0) with a large value for IntAcc (perhaps around 6.0). The default grids
have been chosen such that they provide adequate accuracy at the lowest possible computational cost,
but for all-electron calculations on heavy elements in conjunction with scalar relativistic Hamiltonians
you should examine the grid dependency very carefully and adjust these parameters accordingly to
minimize errors. You should be aware that for large molecules the exchange-correlation integration is
usually not the dominating factor (not even in combination with RI-J).
Similarly important is the value of Thresh that will largely determine the tunaround time for direct
SCF calculations. It may be possible to go to values of 106 108 which will result in large speed-ups.
However, the error in the final energy may then be 3 orders of magnitude larger than the cutoff or,
sometimes, your calculation will fail to converge, due to the limited integral accuracy. In general it
will not be possible to converge a direct SCF calculation to better than Thresh (the program will also
not allow this). For higher accuracy values of maybe 1010 1012 may be used with larger molecules
requiring smaller cutoffs. In cases where the SCF is almost converged but then fails to finally converge
(which is very annoying) decreasing Thresh (and possibly switch to NRSCF) may be a good idea. In
general, TCut should be around 0.01Thresh in order to be on the safe side.
DFT calculations have many good features and in many cases they produce reliable results. In particular
if you study organic molecules it is nevertheless a good idea to check on your DFT results using MP2.
MP2 in the form of RI-MP2 is usually affordable and produces reliable results (in particular for weaker
interactions where DFT is less accurate). In case of a large mismatch between the MP2 and DFT
results the alarm rings in many such cases MP2 is the better choice, but in others (e.g. for redox
processes or transition metal systems) it is not. Remember that SCS-MP2 (RI-SCS-MP2) will usually
produce more accurate results than MP2 itself.
Coupled-cluster calculations become more and more feasible and should be used whenever possible.
The LPNO-CCSD, DLPNO-CCSD and DLPNO-CCSD(T) calculations are available for single-point
calculations and provide accurate results. However, a coupled-cluster study does require careful study
of basis set effects because convergence to the basis set limit is very slow. The established basis
set extrapolation schemes may be very helpful here. For open-shell molecules and in particular for
transition metals one cannot be careful enough with the reference. You have to carefully check that
the Hartree-Fock calculation converged to the desired state in order to get coupled-cluster results
that are meaningful. Orbital optimized MP2, CASSCF or DFT orbitals may help but we have often
encountered convergence difficulties in the coupled-cluster equations with such choices.
Generally speaking, CEPA is often better than CCSD and approaches the quality of CCSD(T). It is,
however, also a little less robust than CC methods because of the less rigorous treatment of the single
excitations in relation to electronic relaxation.
Dont forget: Computers dont solve problems people do. Not denying the importance and desire
to obtain accurate numbers: dont forget that in the end it is the molecule and its chemistry or
spectroscopy that we want to learn something about. The fact that you may be able to compute one
or the other number a little more accurate doesnt mean that this helps us understanding the physics
and chemistry of our target system any better. The danger of getting locked into technicalities and
miss the desired insight is real!
10.3 Converging SCF Calculations
Despite all efforts you may still find molecules where SCF convergence is poor. These are almost invariably
related to open-shell situations and the answer is almost always to provide better starting orbitals. Here is
my standard strategy to deal with this (assuming a DFT calculation):
Perform a small basis set (SV) calculation in using the LSD or BP functional and RI approximation with
a cheap auxiliary basis set. Set Convergence=Loose and MaxIter=200 or so. Turn the FinalGrid off
(NoFinalGrid). The key point is to use a large damping factor and damp until the DIIS comes into a
domain of convergence. This is accomplished by SlowConv or even VerySlowConv. If you have an even
more pathological case you may need to set DampFac larger and DampErr smaller than chosen by these
defaults. This calculation is quite crude and may take many cycles to converge. It will however be
rather quick in terms of wall clock time. If the DIIS gets stuck at some error 0.001 or so the SOSCF (or
even better NRSCF) could be put in operation from this point on.
Use the orbitals of this calculation and GuessMode=CMatrix to start a calculation with the target
basis set. In DFT we normally use a pure GGA functional (e.g. BP86). This calculation normally
converges relatively smoothly.
Use the target functional, grid etc. to get the final calculation converged. In many cases this should
converge fairly well now.
Here are a few other things that can be tried:
Try to start from the orbitals of a related closed-shell species. In general closed-shell MO calculations
tend to converge better. You then hope to reach the convergence radius of another converger for the
open-shell case.
Try to start from the orbitals of a more positive cation. Cation calculations tend to converge better.
10.3 Converging SCF Calculations 753
Try to start from a calculation with a smaller basis set. Smaller basis sets converge better. Then
you have the choice of GuessMode=CMatrix or GuessMode=FMatrix which will affect the convergence
behavior.
Use large level shifts. This increases the number of iterations but stabilizes the converger. (shift
shift 0.5 erroff 0 end)
If your calculations converge to, say 104 in the DIIS error and then creep it is usually a good idea
to invoke the second order (NRSCF) converger. Close to the solution, it is incredibly strong and will
usually converge in 1 or 2 macro-iterations. Each of these cost perhaps 57 or maybe even 10 normal
SCF cycles but after the DIIS has brought you into the radius of the convergence of the NR procedure
its convergence properties are excellent.
If you are doing DFT calculations try to start from a Hartree-Fock solution for your molecule. HF
calculations tend to converge somewhat better because they have a larger HOMO-LUMO gap (there
are of course exceptions).
Carefully look at the starting orbitals (Print[P GuessOrb]=1) and see if they make sense for your
molecule. Perhaps you have to reorder them (using Rotate) to obtain smooth convergence.
Most of the time the convergence problems come from unreasonable structures. Did you make sure
that your coordinates are in the correct units (Angstrom or Bohrs?) and have been correctly recognized
as such by the program?
If you have trouble with UHF calculations try ROHF (especially SAHF or CAHF) first and then go to
the UHF calculation.
Fool around with Guess=Hueckel, PAtom or even HCore.
It may sometimes be better to converge to an undesired state and then take the orbitals of this state,
reorder them (using Rotate) and try to converge to the desired state.
Similarly, bad orbitals may be manipulated using the SCF stability analysis (section 9.7) to provide a
new guess.
Try to start the calculation with a large damping factor (DampFac=0.90; or even larger) and specify a
relatively small DIIS error to turn damping off (say DampErr=0.02;). This will increase the number of
cycles but may guide you into a regime were the calculation actually converges.
The advices above mostly apply to Hartree-Fock and DFT. For CASSCF, the available options and
how they can aid to overcome convergence problems are described in the CASSCF manual section.
In many cases modifying the initial guess or adding a level shift will help. Do not hesitate to use
large level-shifts (e.g 2.0 or even 3.0). The manual is accompanied by CASSCF tutorial that goes
through many details of the process including practical advices on convergence. The choice of initial
guess is crucial. Some guesses work better for organic molecules while others excel for transition-metal
complexes. The tutorial therefore discusses various initial guess options available in ORCA.
If nothing else helps, stop, grab a friend and go to the next pub (you can also send me an unfriendly
e-mail but this will likely not make your calculation converge any quicker; ^).

10.4 Choice of Theoretical Method
The array of available functionals makes it perhaps difficult to decide which one should be used. While this
is a matter of ongoing research and, in the end, can only be answered by experimentation and comparison to
experimental results or high-level ab initio calculations, I may attempt to give some guidelines.
The simplest density functionals (and in general the least accurate) are the local functionals (Functional=LSD).
Although several variants of the local DFT exist in ORCA there is little to choose among them they give
more or less the same result.
The gradient corrected functionals are (very slightly) more expensive because the gradient of the electron
density at each point in space must be computed but they are also significantly more accurate for structures
and energetics of molecules. The various gradient corrected functionals (GGA functionals) are generally
similar in their behavior. The BP functional is probably the most widely used in transition metal chemistry.
The BLYP, PBE or PW91 functionals may also be considered. PWP has been shown to be rather good for
hyperfine coupling predictions of light nuclei in radicals. In addition, since no Hartree-Fock exchange is used
you have the ability to speed up the calculation by a factor of 440 if the RI approximation is employed.
This approximation is really advisable for the LSD and GGA functionals since it leads to very little or no
loss in accuracy while giving large speedups. It is, in fact, automatically chosen to be operative when you use
pure functionals.
In addition, meta-GGAs (TPSS) are available in ORCA and may provide superior results for certain properties
compared to standard GGAs. They are somewhat but not much more expensive to evaluate than standard
GGAs.
For many properties (but not necessarily for geometries), more accurate results are usually given by the
hybrid density functionals that incorporate part of the HF exchange. The computational effort for these is
higher than for the GGA functionals because the HF exchange needs to be computed exactly. Very large
speedups result if this is done via the RIJCOSX approximation. Nevertheless for energetics, properties and
for predictions of charge and spin densities the hybrids appear to be the best choice. The prototype functional
of this kind is B3LYP, which has been very widely used throughout chemistry and is successful for a wide
range of molecular properties. Other hybrids have been less well tested but maybe a good choice in specific
situations, for example the PBE0 functional has been advertised for NMR chemical shift predictions and
other properties. From my personal experience I can also recommend PBE0 and PWP1 as two functionals
that give good predictions for EPR g-values and hyperfine couplings. The TPSSh meta-GGA hybrid is also
very succesful in this area.1
Together with DFT, it is often observed that the atom-pairwise dispersion correction of Stefan Grimme
substantially improves the results at no extra cost.
Dont forget that in present days the MP2 method becomes affordable for molecules of significant size
and there are quite a number of instances where MP2 (and particularly SCS-MP2) will do significantly
1
Some researchers like to adjust the amount of Hartree-Fock exchange according to their needs or what they think
is better than the standard. This increases the semiempirical character of the calculations and may represent
fixes that only work for a given class of compounds and/or properties while worsening the results for others. With
this caveat in mind it is one of the things that you are free to try if you like it. However, we do not recommend it
since it will deteriorate the comparability of your results with those of other workers the vast majority of which
use standard functionals. An alternative to changing the amount of HF exchange could be to simply construct a
linear regression for a number of known cases and then use linear regression.
10.4 Choice of Theoretical Method 755
better than DFT even if it takes a little longer (the RI approximation is also highly recommended here).
The perturbatively corrected functionals (B2PLYP) may also be a very good choice for many problems
(at comparable cost to MP2; note that even for large molecules with more than 1000 basis functions the
MP2 correction only takes about 10-20% of the time required for the preceeding SCF calculation if the RI
approximation is invoked).
Beyond DFT and (SCS)MP2 there are coupled-cluster methods and their implementation in ORCA is
efficient. With the local pair natural orbital methods you can even study molecules of substantial size and
with appealing turnaround times.
When to go to multireference methods is a more complicated question. Typically, this will be the case if
multiplets are desired, pure spin functions for systems with several unpaired electrons, in bond breaking
situations or for certain classes of excited states (loosely speaking: whenever there are weakly interacting
electrons in the system). However, whenever you decide to do so, please be aware that this require substantial
insight into the physics and chemistry of the problem at hand. An uneducated use of CASSCF or MRCI/MRPT
method likely yields numbers that are nonsensical and that at tremendous computational cost. Here, there is
no substitute for experience (and patience ^).
756
11 Publications Related to ORCA
The generic reference for ORCA is:
Neese, F. (2012) The ORCA program system, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2, 7378.
Please do not only cite the above generic reference, but also cite in addition the original papers that report
the development and ORCA implementation of the methods you have used in your studies! The following
publications describe functionality implemented in ORCA. We would highly appreciate if you cite them when
you use the program.
MCD
1. Ye, S. F.; Kupper, C.; Meyer, S.; Andris, E.; Navratil, R.; Krahe, O.; Mondal, B.; Atanasov, M.; Bill,
E.; Roithova, J.; Meyer, F.; Neese, F. (2016) Magnetic Circular Dichroism Evidence for an Unusual
Electronic Structure of a Tetracarbene-Oxoiron(IV) Complex, J. Am. Chem. Soc., 138, 14312-14325.
2. Ye, S. F.; Xue, G. Q.; Krivokapic, I.; Petrenko, T.; Bill, E.; Que, L.; Neese, F. (2015) Magnetic circular
dichroism and computational study of mononuclear and dinuclear iron(IV) complexes, Chem. Sci., 6,
2909-2921.
CASSCF/NEVPT2/MRCI & Magnetism
3. Werncke, C. G.; Suturina, E.; Bunting, P. C.; Vendier, L.; Long, J. R.; Atanasov, M.; Neese, F.; Sabo-
Etienne, S.; Bontemps, S. (2016) Homoleptic Two-Coordinate Silylamido Complexes of Chromium(I),
Manganese(I), and Cobalt(I), Chem. Eur. J., 22, 1668-1674.
4. Rechkemmer, Y.; Breitgoff, F. D.; van der Meer, M.; Atanasov, M.; Hakl, M.; Orlita, M.; Neugebauer,
P.; Neese, F.; Sarkar, B.; van Slageren, J. (2016) A four-coordinate cobalt(II) single-ion magnet with
coercivity and a very high energy barrier, Nat. Commun., 7, 10467.
5. Aravena, D.; Atanasov, M.; Neese, F. (2016) Periodic Trends in Lanthanide Compounds through the
Eyes of Multireference ab Initio Theory, Inorg. Chem., 55, 4457-4469.
6. Suturina, E. A.; Maganas, D.; Bill, E.; Atanasov, M.; Neese, F. (2015) Magneto-Structural Correlations
in a Series of Pseudotetrahedral Co-II(XR)(4) (2-) Single Molecule Magnets: An ab Initio Ligand Field
Study, Inorg. Chem., 54, 9948-9961.
7. Schweinfurth, D.; Sommer, M. G.; Atanasov, M.; Demeshko, S.; Hohloch, S.; Meyer, F.; Neese, F.;
Sarkar, B. (2015) The Ligand Field of the Azido Ligand: Insights into Bonding Parameters and
Magnetic Anisotropy in a Co(II)-Azido Complex, J. Am. Chem. Soc., 137, 1993-2005.
MDCI
757
8. Veis, L.; Antalik, A.; Brabec, J.; Neese, F.; Legeza, O.; Pittner, J. (2016) Coupled Cluster Method
with Single and Double Excitations Tailored by Matrix Product State Wave Functions, J. Phys. Chem.
Lett., 7, 4072-4078.
9. Dutta, A. K.; Neese, F.; Izsak, R. (2016) Speeding up equation of motion coupled cluster theory with
the chain of spheres approximation, J. Chem. Phys., 144, 034102.
DLPNO
10. Schneider, W. B.; Bistoni, G.; Sparta, M.; Saitow, M.; Riplinger, C.; Auer, A. A.; Neese, F. (2016)
Decomposition of Intermolecular Interaction Energies within the Local Pair Natural Orbital Coupled
Cluster Framework, J. Chem. Theory Comput., 12, 4778-4792.
11. Riplinger, C.; Pinski, P.; Becker, U.; Valeev, E. F.; Neese, F. (2016) Sparse maps-A systematic
infrastructure for reduced-scaling electronic structure methods. II. Linear scaling domain based pair
natural orbital coupled cluster theory, J. Chem. Phys., 144, 024109.
12. Pavosevic, F.; Pinski, P.; Riplinger, C.; Neese, F.; Valeev, E. F. (2016) SparseMaps-A systematic in-
frastructure for reduced-scaling electronic structure methods. IV. Linear-scaling second-order explicitly
correlated energy with pair natural orbitals, J. Chem. Phys., 144, 144109.
13. Kubas, A.; Berger, D.; Oberhofer, H.; Maganas, D.; Reuter, K.; Neese, F. (2016) Surface Adsorption
Energetics Studied with Gold Standard Wave Function-Based Ab Initio Methods: Small-Molecule
Binding to TiO2(110), J. Phys. Chem. Lett., 7, 4207-4212.
14. Isegawa, M.; Neese, F.; Pantazis, D. A. (2016) Ionization Energies and Aqueous Redox Potentials
of Organic Molecules: Comparison of DFT, Correlated ab Initio Theory and Pair Natural Orbital
Approaches, J. Chem. Theory Comput., 12, 2272-2284.
15. Guo, Y.; Sivalingam, K.; Valeev, E. F.; Neese, F. (2016) SparseMaps-A systematic infrastructure for
reduced-scaling electronic structure methods. III. Linear-scaling multireference domain-based pair
natural orbital N-electron valence perturbation theory, J. Chem. Phys., 144, 094111.
16. Dutta, A. K.; Neese, F.; Izsak, R. (2016) Towards a pair natural orbital coupled cluster method for
excited states, J. Chem. Phys., 145, 034102.
17. Datta, D.; Kossmann, S.; Neese, F. (2016) Analytic energy derivatives for the calculation of the
first-order molecular properties using the domain-based local pair-natural orbital coupled-cluster theory,
J. Chem. Phys., 145, 114101.
18. Pinski, P.; Riplinger, C.; Valeev, E. F.; Neese, F. (2016) Sparse maps-A systematic infrastructure
for reduced-scaling electronic structure methods. I. An efficient and simple linear scaling local MP2
method that uses an intermediate basis of pair natural orbitals, J. Chem. Phys., 143, 034108.
19. Mondal, B.; Neese, F.; Ye, S. F. (2015) Control in the Rate-Determining Step Provides a Promising
Strategy To Develop New Catalysts for CO2 Hydrogenation: A Local Pair Natural Orbital Coupled
Cluster Theory Study, Inorg. Chem., 54, 7192-7198.
20. Liakos, D. G.; Sparta, M.; Kesharwani, M. K.; Martin, J. M. L.; Neese, F. (2015) Exploring the
Accuracy Limits of Local Pair Natural Orbital Coupled-Cluster Theory, J. Chem. Theory Comput., 11,
1525-1539.
758 11 Publications Related to ORCA
21. Liakos, D. G.; Neese, F. (2015) Domain Based Pair Natural Orbital Coupled Cluster Studies on Linear
and Folded Alkane Chains, J. Chem. Theory Comput., 11, 2137-2143.
22. Liakos, D. G.; Neese, F. (2015) Is It Possible To Obtain Coupled Cluster Quality Energies at near
Density Functional Theory Cost? Domain-Based Local Pair Natural Orbital Coupled Cluster vs
Modern Density Functional Theory, J. Chem. Theory Comput., 11, 4054-4063.
23. Demel, O.; Pittner, J.; Neese, F. (2015) A Local Pair Natural Orbital-Based Multireference Mukherjees
Coupled Cluster Method, J. Chem. Theory Comput., 11, 3104-3114.
F12
24. Pavosevic, F.; Pinski, P.; Riplinger, C.; Neese, F.; Valeev, E. F. (2016) SparseMaps-A systematic in-
frastructure for reduced-scaling electronic structure methods. IV. Linear-scaling second-order explicitly
correlated energy with pair natural orbitals, J. Chem. Phys., 144, 144109.
XAS/XES
25. Van Kuiken, B. E.; Hahn, A. W.; Maganas, D.; DeBeer, S. (2016) Measuring Spin-Allowed and
Spin-Forbidden d-d Excitations in Vanadium Complexes with 2p3d Resonant Inelastic X-ray Scattering,
Inorg. Chem., 55, 11497-11501.
26. Rees, J. A.; Wandzilak, A.; Maganas, D.; Wurster, N. I. C.; Hugenbruch, S.; Kowalska, J. K.; Pollock,
C. J.; Lima, F. A.; Finkelstein, K. D.; DeBeer, S. (2016) Experimental and theoretical correlations
between vanadium K-edge X-ray absorption and K emission spectra, J. Biol. Inorg. Chem., 21,
793-805.
27. Martin-Diaconescu, V.; Chacon, K. N.; Delgado-Jaime, M. U.; Sokaras, D.; Weng, T. C.; DeBeer, S.;
Blackburn, N. J. (2016) K beta Valence to Core X-ray Emission Studies of Cu(I) Binding Proteins
with Mixed Methionine - Histidine Coordination. Relevance to the Reactivity of the M- and H-sites of
Peptidylglycine Monooxygenase, Inorg. Chem., 55, 3431-3439.
28. Maganas, D.; Trunschke, A.; Schlogl, R.; Neese, F. (2016) A unified view on heterogeneous and
homogeneous catalysts through a combination of spectroscopy and quantum chemistry, Farad. Discuss.,
188, 181-197.
29. Kowalska, J. K.; Hahn, A. W.; Albers, A.; Schiewer, C. E.; Bjornsson, R.; Lima, F. A.; Meyer,
F.; DeBeer, S. (2016) X-ray Absorption and Emission Spectroscopic Studies of L2Fe2S2 (n) Model
Complexes: Implications for the Experimental Evaluation of Redox States in Iron-Sulfur Clusters,
Inorg. Chem., 55, 4485-4497.
30. Rees, J. A.; Martin-Diaconescu, V.; Kovacs, J. A.; DeBeer, S. (2015) X-ray Absorption and Emission
Study of Dioxygen Activation by a Small-Molecule Manganese Complex, Inorg. Chem., 54, 6410-6422.
31. Rees, J. A.; Bjornsson, R.; Schlesier, J.; Sippel, D.; Einsle, O.; DeBeer, S. (2015) The Fe-V Cofactor of
Vanadium Nitrogenase Contains an Interstitial Carbon Atom, Angew. Chem. Int. Ed., 54, 13249-13252.
Resonance Raman
759
32. Maganas, D.; Trunschke, A.; Schlogl, R.; Neese, F. (2016) A unified view on heterogeneous and
homogeneous catalysts through a combination of spectroscopy and quantum chemistry, Farad. Discuss.,
188, 181-197.
QM/MM
33. Rokhsana, D.; Large, T. A. G.; Dienst, M. C.; Retegan, M.; Neese, F. (2016) A realistic in silico model
for structure/function studies of molybdenum-copper CO dehydrogenase Journal of Biological, Inorg.
Chem., 21, 491-499.
34. Retegan, M.; Krewald, V.; Mamedov, F.; Neese, F.; Lubitz, W.; Cox, N.; Pantazis, D. A. (2016) A
five-coordinate Mn(IV) intermediate in biological water oxidation: spectroscopic signature and a pivot
mechanism for water binding, Chem. Sci., 7, 72-84.
35. Sundararajan, M.; Neese, F. Distal (2015) Histidine Modulates the Unusual O-Binding of Nitrite to
Myoglobin: Evidence from the Quantum Chemical Analysis of EPR Parameters, Inorg. Chem., 54,
7209-7217.
NRVS
36. Ogata, H.; Kramer, T.; Wang, H. X.; Schilter, D.; Pelmenschikov, V.; van Gastel, M.; Neese, F.;
Rauchfuss, T. B.; Gee, L. B.; Scott, A. D.; Yoda, Y.; Tanaka, Y.; Lubitz, W.; Cramer, S. P. (2015)
Hydride bridge in NiFe -hydrogenase observed by nuclear resonance vibrational spectroscopy, Nat.
Commun., 6, 7890.
Applications
37. Mondal, B.; Neese, F.; Ye, S. F. (2016) Toward Rational Design of 3d Transition Metal Catalysts
for CO2 Hydrogenation Based on Insights into Hydricity-Controlled Rate-Determining Steps, Inorg.
Chem., 55, 5438-5444.
38. Mondal, B.; Neese, F.; Ye, S. F. (2015) Control in the Rate-Determining Step Provides a Promising
Strategy To Develop New Catalysts for CO2 Hydrogenation: A Local Pair Natural Orbital Coupled
Cluster Theory Study, Inorg. Chem., 54, 7192-7198.
39. Krewald, V.; Retegan, M.; Cox, N.; Messinger, J.; Lubitz, W.; DeBeer, S.; Neese, F.; Pantazis, D. A.
(2015) Metal oxidation states in biological water splitting, Chem. Sci., 6, 1676-1695.
40. Kochem, A.; Weyhermuller, T.; Neese, F.; van Gastel, M. (2015) EPR and Quantum Chemical
Investigation of a Bioinspired Hydrogenase Model with a Redox-Active Ligand in the First Coordination
Sphere, Organometallics, 34, 995-1000.
41. Kochem, A.; Bill, E.; Neese, F.; van Gastel, M. (2015) Mossbauer and computational investigation of
a functional NiFe hydrogenase model complex, Chem. Comm., 51, 2099-2102.
COSX
42. Dutta, A. K.; Neese, F.; Izsak, R. (2016) Speeding up equation of motion coupled cluster theory with
the chain of spheres approximation, J. Chem. Phys., 144, 034102.
43. Christian, G. J.; Neese, F.; Ye, S. F. (2016) Unravelling the Molecular Origin of the Regiospecificity
in Extradiol Catechol Dioxygenases, Inorg. Chem., 55, 3853-3864.
SARC
44. Aravena, D.; Neese, F.; Pantazis, D. A. (2016) Improved Segmented All-Electron Relativistically
Contracted Basis Sets for the Lanthanides, J. Chem. Theory Comput., 12, 1148-1156.
Review
45. Atanasov, M.; Aravena, D.; Suturina, E.; Bill, E.; Maganas, D.; Neese, F. (2015) First principles
approach to the electronic structure, magnetic anisotropy and spin relaxation in mononuclear 3d-
transition metal single molecule magnets, Coord. Chem. Rev., 289, 177-214.
Multireference CI Module and its application to EPR properties and optical spectra:
46. Demel, O.; Pittner, J.; Neese, F. (2015) A Local Pair Natural Orbital-Based Multireference Mukherjees
Coupled Cluster Method, J. Chem. Theory Comput., 11, 3104.
47. Nooijen, M.; Demel, O.; Datta, D.; Kong, L.; Shamasundar K. R.; Lotrich, V.; Huntington, L. M.;
Neese, F. (2014) , J. Chem. Phys., 140, 081102.
48. Ganyushin, D.; Neese, F. (2013) A fully variational spin-orbit coupled complete active space self-
consistent field approach: Application to electron paramagnetic resonance g-tensors, J. Chem. Phys.,
138, 104113.
49. Atanasov, M.; Zadrozny, J. M.; Long, J. R.; Neese, F. (2013) A theoretical analysis of chemical
bonding, vibronic coupling, and magnetic anisotropy in linear iron(II) complexes with single-molecule
magnet behavior, Chem. Sci., 4, 139156.
50. Atanasov, M.; Ganyushin, D.; Pantazis, D. A.; Sivalingam, K.; Neese, F. (2011) Detailed Ab Initio
First-Principles Study of the Magnetic Anisotropy in a Family of Trigonal Pyramidal Iron(II) Pyrrolide
Complexes, Inorg. Chem, 50, 74607477.
51. Neese, F.; Pantazis, D. A. (2011) What is not required to make a single molecule magnet, Faraday
Discussions, 148, 229238.
52. Duboc, C.; Ganyushin, D.; Sivalingam, K.; Collomb, M. N.; Neese, F. (2010) Systematic Theoretical
Study of the Zero-Field Splitting in Coordination Complexes of Mn(III). Density Functional Theory
versus Multireference Wave Function Approaches, J. Phys. Chem. A, 114, 1075010758.
53. Sundararajan, M.; Ganyushin, D.; Ye, S.; Neese, F. (2009) Multireference ab initio studies of Zero-Field
Splitting and Magnetic Circular Dichroism Spectra of Tetrahedral Co(II) Complexes, Dalton Trans.,
30, 60216036.
54. Ganyushin, D.; Neese, F. (2008) First principles calculation of magnetic circular dichroism spectra, J.
Chem. Phys., 128, 114117.
55. Petrenko, T.; Neese, F. (2007) A general efficient quantum chemical method for predicting absorption
bandshapes, resonance Raman spectra and excitation profiles for larger molecules. J. Chem. Phys.,
127, 164319.
761
56. Neese, F. (2007) Analytic Derivative Calculation of Electronic g-Tensors based on Multireference
Configuration Interaction Wavefunctions. Mol. Phys. (honorary issue for Prof. Peter Pulay), 105,
25072514.
57. Neese, F.; Petrenko, T.; Ganyushin, D.; Olbrich, G. (2007) Advanced Aspects of ab initio Theoretical
Spectroscopy of Open-Shell Transition Metal Ions. Coord. Chem. Rev., 205, 288327.
58. Neese, F. (2006) Importance of Direct Spin-Spin Coupling and Spin-Flip Excitations for the Zero-Field
Splittings of Transition Metal Complexes: A Case Study, J. Am. Chem. Soc., 128, 1021310222.
59. Chalupsk y, J.; Neese, F.; Solomon, E.I.; Ryde, U.; Rulsek, L. (2006) Identification of intermediates in
the reaction cycle of multicopper oxidases by quantum chemical calculations of spectroscopic parameters,
Inorg. Chem., 45, 1105111059.
60. Neese, F. (2006) Theoretical spectroscopy of model-nonheme [FeIV O(NH3 )5 ]2+ complexes with triplet
and quintet ground states using multireference ab initio and density functional theory methods. J.
Inorg. Biochem. (special issue on high-valent Fe(IV)), 716726.
61. Ganyushin, D.; Neese, F. (2006) First Principle Calculation of Zero-Field Splittings, J. Chem. Phys.,
125, 024103.
62. Ray, K.; Weyherm uller, T.; Neese, F.; Wieghardt, K. (2005) Electronic Structure of Square-Planar
Bis(benzene-1,2-dithiolate)metal Complexes [M(L)2 ]z (z = 2, 1, 0; M = Ni, Pd, Pt, Cu, Au): An
experimental, Density Functional and Correlated ab initio Study. Inorg. Chem., 44, 53455360.
63. Schoneboom, J.; Neese, F.; Thiel, W. (2005) Towards Identification of the Compound I Reactive
Intermediate in Cytochrome P450 Chemistry: A QM/MM Study of its EPR and Mossbauer Parameters,
J. Am. Chem. Soc., 127, 58405853.
64. Wanko, M.; Hoffmann, M.; Strodel, P.; Thiel, W.; Neese, F.; Frauenheim, T.; Elstner, M. (2005)
Calculating Absorption Shifts for Retinal Proteins: Computational Challenges J. Phys. Chem. B, 109,
36063615.
65. Neese, F. (2004) Sum Over States Based Multireference ab initio Calculation of EPR Spin Hamiltonian
Parameters for Transition Metal Complexes. A Case Study Mag. Res. Chem., 42, S187S198.
66. Neese, F. (2003) Correlated ab Initio Calculation of Electronic g-Tensors Using a Sum Over States
Formulation. Chem. Phys. Lett., 380/56, 721728.
67. Neese, F. (2003) A Spectroscopy Oriented Configuration Interaction Procedure, J. Chem. Phys., 119,
94289443.
68. Neese, F. (2001) Configuration Interaction Calculation of Electronic g-Tensors in Transition Metal
Complexes, Int. J. Quant. Chem., 83, 104114.
69. Neese, F.; Solomon, E.I. (1998) Calculation of Zero-Field Splittings, g-values and the Relativistic
Nephelauxetic Effect in Transition Metal Complexes. Application to High Spin Ferric Complexes.
Inorg. Chem., 37, 65686582.
Internally Contracted Multireference CI:

70. Sivalingam, K.; Krupicka, M.; Auer, A. A.; Neese, F. (2016) Comparison of fully internally and
strongly contracted multireference configuration interaction procedures. J. Chem. Phys., 145, 54104.
Ab initio Ligand Field Analyis:
71. Atanasov, M.; Ganyushin, D.; Sivalingam, K.; Neese, F. )in Molecular Electronic Structures of
Transition Metal Complexes II (eds. Mingos, D. M. P., Day, P. Dahl, J. P.) 149220 (Springer Berlin
Heidelberg, 2011).
Pair Natural Orbital Local Correlation Methods:
72. Riplinger, C.; Sandhoefer B.; Hansen, A.; Neese, F. (2013) Natural triple excitations in local coupled
cluster calculations with pair natural orbitals, J. Chem. Phys., 139, 134101.
73. Riplinger, C.; Neese, F. (2013) An efficient and near linear scaling pair natural orbital based local
coupled cluster method, J. Chem. Phys., 138, 034106.
74. Liakos, D. G.; Neese, F. (2012) Improved Correlation Energy Extrapolation Schemes Based on Local
Pair Natural Orbital Methods, J. Phys. Chem. A, 116, 48014816.
75. Huntington, L. M. J.; Hansen, A.; Neese, F.; Nooijen, M. (2012) Accurate thermochemistry from
a parameterized coupled-cluster singles and doubles model and a local pair natural orbital based
implementation for applications to larger systems, J. Chem. Phys., 136, 064101.
76. Izsak, R.; Hansen, A.; Neese, F. (2012) The resolution of identity and chain of spheres approximations
for the LPNO-CCSD singles Fock term, Mol. Phys., 110, 24132417.
77. Liakos, D. G.; Hansen, A.; Neese, F. (2011) Weak Molecular Interactions Studied with Parallel
Implementations of the Local Pair Natural Orbital Coupled Pair and Coupled Cluster Methods, J.
Chem. Theory Comput., 7, 7687.
78. Hansen, A.; Liakos, D. G.; Neese, F. (2011) Efficient and accurate local single reference correlation
methods for high-spin open-shell molecules using pair natural orbitals, J. Chem. Phys., 135, 214102.
79. Kollmar, C.; Neese, F. (2011) An orbital-invariant and strictly size extensive post-Hartree-Fock
correlation functional, J. Chem. Phys., 135, 084102.
80. Kollmar, C.; Neese, F. (2011) The relationship between double excitation amplitudes and Z vector
components in some post-Hartree-Fock correlation methods, J. Chem. Phys., 135, 064103.
81. Neese, F.; Liakos, D.; Hansen, A. (2009) Efficient and accurate local approximations to the coupled
cluster singles and doubles method using a truncated pair natural orbital basis J. Chem. Phys., 131,
064103.
82. Neese, F.; Wennmohs, F.; Hansen, A. (2009) Efficient and accurate local approximations to coupled
electron pair approaches. An attempt to revive the pair-natural orbital method J. Chem. Phys., 130,
114108.
Coupled-Cluster and Coupled Pair Implementation (MDCI module):

763
83. Pavosevic, F.; Neese, F.; Valeev, E. F. (2014) Geminal-spanning orbitals make explicitly correlated
reduced-scaling coupled-cluster methods robust, yet simple, J. Chem. Phys., 141, 054106
84. Kollmar, C.; Neese, F. (2010) The coupled electron pair approximation: variational formulation and
spin adaptation, Mol. Phys., 108, 24492458.
85. Neese, F.; Wennmohs, F.; Hansen, A.; Grimme, S. (2009) Accurate Theoretical Chemistry with
Coupled Electron Pair Models Acc. Chem. Res., 42(5), 641648.
86. Wennmohs, F.; Neese, F. (2008) A Comparative Study of Single Reference Correlation Methods of the
Coupled-Pair Type, Chem. Phys. (70th birthday issue for Prof. Peyerimhoff), 343, 217230.
Explicit Correlation (R12, F12 Methods):
87. Liakos D. G.; Izsak R.; F.; Valeev, E. F.; Neese, F. (2013) What is the most efficient way to reach the
canonical MP2 basis set limit?, Mol. Phys., 111, 2653
ANO basis sets:
88. Neese, F.; Valeev, E. F. (2011) Revisiting the Atomic Natural Orbital Approach for Basis Sets: Robust
Systematic Basis Sets for Explicitly Correlated and Conventional Correlated ab initio Methods?, J.
Orbital Optimized MP2:
89. Sandhoefer, B.; Kossmann, S.; Neese, F. (2013) Derivation and assessment of relativistic hyperfine-
coupling tensors on the basis of orbital-optimized second-order Moller-Plesset perturbation theory and
the second-order DouglasKrollHess transformation, J. Chem. Phys., 138, 104102.
90. Kossmann, S.; Neese, F. (2010) Correlated ab Initio Spin Densities for Larger Molecules: Orbital-
Optimized Spin-Component-Scaled MP2 Method, J. Phys. Chem. A, 114, 11768-11781.
91. Neese, F.; Schwabe, T.; Kossmann, S.; Schirmer, B.; Grimme, S. (2009) Assessment of Orbital Opti-
mized, Spin-Component Scaled Second Order Many Body Perturbation Theory for Thermochemistry
and Kinetics. J. Chem. Theory Comput., 5, 3060-3073.
Analytic Hessian Implementation:
92. Bykov, D.; Petrenko, T.; Izsak, R.; Kossmann, S.; Becker, U.; Valeev, E.; Neese, F. (2015) Efficient
implementation of the analytic second derivatives of Hartree-Fock and hybrid DFT energies: a detailed
analysis of different approximations, Mol. Phys., 113, 1961.
The Split-J, Split-RI-J, RIJCOSX and RI-JK methods:
93. Izsak, R.; Neese, F. (2013) Speeding up spin-component-scaled third-order pertubation theory with
the chain of spheres approximation: the COSX-SCS-MP3 method, Mol. Phys., 111, 1190.
94. Izsak, R.; Neese, F. (2011) An overlap fitted chain of spheres exchange method, J. Chem. Phys., 135,
144105.
95. Kossmann, S.; Neese, F. (2010) Efficient Structure Optimization with Second-Order Many-Body
Perturbation Theory: The RIJCOSX-MP2 Method, J. Chem. Theory Comput., 6, 2325-2338.
96. Kossmann, S.; Neese, F. (2009) Comparison of Two Efficient Approximate HartreeFock Approaches.
Chem. Phys. Lett., 481, 240-243.
97. Neese, F.; Wennmohs, F.; Hansen, A.; Becker, U. (2009) Efficient, approximate and parallel Hartree
Fock and hybrid DFT calculations. A chain-of-spheres algorithm for the HartreeFock exchange,
Chem. Phys., 356, 98109.
98. Neese, F. (2003) An Improvement of the Resolution of the Identity Approximation for the Calculation
of the Coulomb Matrix, J. Comp. Chem., 24, 17401747.
DFT/HartreeFock Theory of EPR Parameters:
99. Sandhoefer, B.; Neese, F. (2012) One-electron contributions to the g-tensor for second-order Douglas
KrollHess theory, J. Chem. Phys., 137, 094102.
100. Ganyushin, D.; Gilka, N.; Taylor, P. R.; Marian, C. M.; Neese, F. (2010) The resolution of the identity
approximation for calculations of spin-spin contribution to zero-field splitting parameters, J. Chem.
Phys., 132, 144111.
101. Duboc, C.; Ganyushin, D.; Sivalingam, K.; Collomb, M. N.; Neese, F. (2010) Systematic Theoretical
Study of the Zero-Field Splitting in Coordination Complexes of Mn(III). Density Functional Theory
versus Multireference Wave Function Approaches, J. Phys. Chem. A, 114, 1075010758.
102. Neese, F. (2009) First principles approach to Spin-Hamiltonian Parameters, invited chapter in Misra,
S.K. Multifrequency EPR: Theory and Applications, Wiley-VCH, pp. 297326.
103. Pantazis, D. A.; Orio, M.; Petrenko, T.; Messinger, J.; Lubitz, W.; Neese, F. (2009) A new quantum
chemical approach to the magnetic properties of oligonuclear transition metal clusters: Application to
a model for the tetranuclear manganese cluster of Photosystem II Chem. Eur. J., 15(20), 51085123.
104. Riplinger, C.; Kao, J.P.Y.; Rosen, G.M.; Kathirvelu, V.; Eaton, G.R.; Eaton, S.S.; Kutateladze, A.;
Neese F. (2009) Interaction of Radical Pairs Through-Bond and Through-Space: Scope and Limitations
of the Point-Dipole Approximation in Electron Paramagnetic Resonance Spectroscopy, J. Am. Chem.
Soc., 131, 1009210106.
105. Cirera, J.; Ruiz, E.; Alvarez, S.; Neese, F.; Kortus, J. (2009) How to Build Molecules with Large
Magnetic Anisotropy. Chem. Eur. J., 15(16), 40784087.
106. Neese, F. (2008) Spin Hamiltonian Parameters from First Principle Calculations: Theory and Applica-
tion. In Hanson, G.; Berliner, L. (Eds), Biological Magnetic Resonance, pp. 175232.
107. Komann, S.; Kirchner, B.; Neese, F. (2007) Performance of modern density functional theory for the
prediction of hyperfine structure: meta-GGA and double hybrid functionals, Molec. Phys. (Arthur
Schweiger memorial issue), 105, 20492071.
108. Neese, F. (2007) Calculation of the Zero-Field Splitting Tensor Using Hybrid Density Functional and
Hartree-Fock Theory. J. Chem. Phys., 127, 164112.
765
109. Neese, F. (2006) Importance of Direct Spin-Spin Coupling and Spin-Flip Excitations for the Zero-Field
Splittings of Transition Metal Complexes: A Case Study, J. Am. Chem. Soc., 128, 1021310222.
110. Sinnecker, S.; Neese, F. (2006) Spin-Spin Contributions to the Zero-Field Splitting Tensor in Organic
Triplets, Carbenes and Biradicals A Density Functional and ab initio Study. J. Phys. Chem. A, 110,
1226712275.
111. Sinnecker, S.; Rajendran, A.; Klamt, A.; Diedenhofen, M.; Neese, F. (2006) Calculation of Solvent Shifts
on Electronic G-Tensors with the Conductor-Like Screening Model (COSMO) and its Self-Consistent
Generalization to Real Solvents (COSMO-RS), J. Phys. Chem. A, 110, 22352245.
112. Neese, F.; Wolf, A.; Reiher, M.; Fleig, T.; Hess, B.A. (2005) Higher Order Douglas-Kroll Calculation
of Electric Field Gradients, J. Chem. Phys., 122, 204107.
113. Neese, F. (2005) Efficient and Accurate Approximations to the Molecular Spin-Orbit Coupling Operator
and their use in Molecular g-Tensor Calculations, J. Chem. Phys., 122, 034107.
114. Ray, K.; Begum, A; Weyherm uller, T.; Piligkos, S.; van Slageren, J.; Neese, F.; Wieghardt, K. (2005)
The Electronic Structure of the Isoelectronic, Square Planar Complexes [FeII (L)2 ]2 and [CoIII (LBu )2 ]
(L2 and (LBu )2 = benzene-1,2-dithiolates): an Experimental and Density Functional Theoretical
Study, J. Am. Chem. Soc., 127, 44034415.
115. Neese, F. (2003) Metal and Ligand Hyperfine Couplings in Transition Metal Complexes. The Effect
of Spin-Orbit Coupling as Studied by Coupled Perturbed Kohn-Sham Theory and Hybrid Density
Functionals, J. Chem. Phys., 117, 39393948.
116. Neese, F. (2001) Prediction of Electron Paramagnetic Resonance g-values by Coupled Perturbed
Hartree-Fock and Kohn-Sham Theory. J. Chem. Phys., 115, 1108011096.
117. Neese, F. (2001) Theoretical Study of Ligand Superhyperfine Structure. Application to Cu(II)
Complexes. J. Phys. Chem. A, 105, 42904299.
Corresponding Orbital Transformation:
118. Neese, F. (2004) Definition of Corresponding Orbitals and the Diradical Character in Broken Symmetry
DFT Calculations on Spin Coupled Systems. J. Phys. Chem. Solids, 65, 781785.
Excited States Methods and Resonance Raman Spectra:
119. Schapiro, I.; Sivalingam K.; Neese, F. (2015) Assessment of n-Electron Valence State Perturbation
Theory for Vertical Excitation Energies, J. Chem. Theory Comput., 9, 3567.
120. Roemelt, M.; Neese, F. (2013) Excited States of Large Open-Shell Molecules: An Efficient, General,
and Spin-Adapted Approach Based on a Restricted Open-Shell Ground State Wave function, J. Phys.
Chem. A, 117, 3069.
121. Petrenko, T.; Kossmann, S.; Neese, F. (2011) Efficient time-dependent density functional theory
approximations for hybrid density functionals: Analytical gradients and parallelization, J. Chem. Phys.,
134, 054116.
122. Petrenko, T.; Krylova, O.; Neese, F.; Sokolowski, M. (2009) Optical Absorption and Emission
Properties of Rubrene: Insight by a Combined Experimental and Theoretical Study. New J. Phys., 11,
015001.
123. Grimme, S.; Neese, F. (2007) Double Hybrid Density Functional Theory for Excited States of Molecules,
J. Chem. Phys., 127, 154116.
124. Petrenko, T.; Ray, K.; Wieghardt, K.; Neese, F. (2006) Vibrational Markers for the Open-Shell
Character of Metal bis-Dithiolenes: An Infrared, resonance Raman and Quantum Chemical Study. J.
Am. Chem. Soc., 128, 44224436.
125. Neese, F.; Olbrich, G. (2002) Efficient use of the Resolution of the Identity Approximation in Time-
Dependent Density Functional Calculations with Hybrid Functionals, Chem. Phys. Lett., 362, 170178.
Absorption, Resonance Raman and Fluorescence Spectra:
126. Petrenko, T.; Neese F. (2012) Efficient and automatic calculation of optical band shapes and resonance
Raman spectra for larger molecules within the independent mode displaced harmonic oscillator model,
J. Chem. Phys., 137, 234107.
127. Petrenko, T.; Neese, F. (2007) A general efficient quantum chemical method for predicting absorption
bandshapes, resonance Raman spectra and excitation profiles for larger molecules. J. Chem. Phys.,
127, 164319.
128. Petrenko, T.; Krylova, O.; Neese, F. Sokolowski, M. (2009) Optical Absorption and Emission Properties
of Rubrene: Insight by a Combined Experimental and Theoretical Study. New J. Phys., 11, 015001.
Magnetic Circular Dichroism Spectra:
129. Westphal, A.; Broda, H.; Kurz, P.; Neese, F.; Tuczek, F. (2012) Magnetic Circular Dichroism Spectrum
of the Molybdenum(V) Complex (Mo(O)Cl3 dppe): C-Term Signs and Intensities for Multideterminant
Excited Doublet States, Inorg. Chem., 51, 57485763.
130. van Slageren, J.; Piligkos, S.; Neese, F. (2010) Magnetic circular dichroism spectroscopy on the Cr(8)
antiferromagnetic ring, Dalton Trans., 39, 49995004.
131. Sundararajan, M.; Ganyushin, D.; Ye, S.; Neese, F. (2009) Multireference ab initio studies of Zero-Field
Splitting and Magnetic Circular Dichroism Spectra of Tetrahedral Co(II) Complexes, Dalton Trans.,
30, 60216036.
132. Piligkos, S.; Slep, L.; Weyherm

uller, T.; Chaudhuri, P.; Bill, E.; Neese, F. (2009) Magnetic Circular
Dichroism Spectroscopy of weakly exchange coupled dimers. A model study. Coord. Chem. Rev., 253,
23522362.
133. Ganyushin, D.; Neese, F. (2008) First principles calculation of magnetic circular dichroism spectra, J.
Chem. Phys., 128, 114117.
134. Neese, F.; Solomon, E.I. (1999) MCD C-term Signs, Saturation Behavior and Determination of Band
Polarizations in Randomly Oriented Systems with Spin S > 1/2. Applications to S = 1/2 and S = 5/2.
Inorg. Chem., 38, 18471865.
767
M
ossbauer Isomer shifts:
135. R
omelt, M.; Ye, S.; Neese, F. (2009) Calibration of Mossbauer Isomer Shift Calculations for Modern
Density Functional Theory: meta-GGA and Double Hybrid Functionals Inorg. Chem., 48, 784785.
136. Sinnecker, S.; Slep, L.; Bill, E.; Neese, F. (2005) Performance of Nonrelativistic and Quasirelativistic
Hybrid DFT for the Prediction of Electric and Magnetic Hyperfine Parameters in 57 Fe Mossbauer
Spectra, Inorg. Chem., 44, 22452254.
137. Neese, F. (2002) Prediction and Interpretation of Isomer Shifts in 57 Fe Mossbauer Spectra by Density
Functional Theory. Inorg. Chim. Acta (special Karl Wieghardt honorary issue), 337C, 181192.
Nuclear Resonance Vibrational Spectra:
138. Petrenko T,; Sturhahn, W.; Neese, F. (2008) First principles calculation of Nuclear Resonance
Vibrational Spectra, Hyperfine Interactions, 175, 165174.
139. DeBeer-George, S.; Petrenko, T.; Aliaga-Alcade, N.; Bill, E.; Mienert, B.; Sturhan, W.; Ming, Y.;
Wieghardt, K.; Neese, F. (2007) Characterization of a Genuine Iron(V)Nitrido Species by Nuclear
Resonant Vibrational Spectroscopy Coupled to Density Functional Calculations, J. Am. Chem. Soc.,
129, 1105311060.
X-Ray Absorption and X-Ray Emission Spectra:
140. Roemelt, M.; Beckwith, M. A.; Duboc, C.; Collomb, M.-N.; Neese, F.; DeBeer, S. (2012) Manganese
K-Edge X-Ray Absorption Spectroscopy as a Probe of the Metal-Ligand Interactions in Coordination
Compounds, Inorg. Chem., 51, 680687.
141. Chandrasekaran, P.; Stieber, S. C. E.; Collins, T. J.; Que, L.; Neese, F.; DeBeer, S. (2011) Prediction
of high-valent iron K-edge absorption spectra by time-dependent Density Functional Theory, Dalton
Trans., 40, 1107011079.
142. Beckwith, M. A.; Roemelt, M.; Collomb, M. N.; Duboc, C.; Weng, T. C.; Bergmann, U.; Glatzel,
P.; Neese, F.; DeBeer, S. (2011) Manganese K beta X-ray Emission Spectroscopy As a Probe of
Metal-Ligand Interactions, Inorg. Chem, 50, 83978409.
143. Lee, N.; Petrenko, T.; Bergmann, U.; Neese, F.; DeBeer, S. (2010) Probing Valence Orbital Composition
with Iron K beta X-ray Emission Spectroscopy, J. Am. Chem. Soc., 132, 97159727.
144. DeBeer-George, S.; Neese, F. (2010) Calibration of Scalar Relativistic Density Functional Theory for
the Calculation of Sulfur K-Edge X-ray Absorption Spectra, Inorg. Chem, 49, 18491853.
145. DeBeer-George, S.; Petrenko, T.; Neese, F. (2008) Prediction of Iron- K-edge Absorption Spectra
using Time-Dependent Density Functional Theory, J. Phys. Chem. A., 112, 1293612943.
146. DeBeer-George, S.; Petrenko, T.; Neese, F. (2008) Time-dependent density functional calculations of
ligand K-edge X-ray absorption spectra, Inorg. Chim. Acta (60th birthday issue of Prof. E.I. Solomon),
361, 965972.
Double hybrid density functionals:

147. Grimme, S.; Neese, F. (2007) Double Hybrid Density Functional Theory for Excited States of Molecules,
J. Chem. Phys., 127, 154116.
148. Neese, F.; Schwabe, T.; Grimme, S. (2007) Analytic Derivatives for Perturbatively Corrected Double
Hybrid Density Functionals, J. Chem. Phys., 126, 124115.
149. Komann, S.; Kirchner, B.; Neese, F. (2007) Performance of modern density functional theory for
the prediction of hyperfine structure: meta-GGA and double hybrid functionals, Mol. Phys. (Arthur
Schweiger memorial issue), 105, 20492071.
Dispersion Corrections to DFT:
(not originally implemented in ORCA but the ORCA implementation is based on the code described in these
papers)
150. Grimme, S. (2004) J. Comput. Chem., 25, 14631476.
151. Grimme, S. (2006) J. Comput. Chem., 27, 17871799.
152. Grimme, S.; Antony, J. Ehrlich, S. Krieg, H. (2010) J. Chem. Phys., 132, 154104.
153. Grimme, S.; Ehrlich, S.; Goerigk, L. (2011) J. Comput. Chem., 32, 1456-1465.
gCP correction to HF and DFT:
(ORCA implementation is based upon the code used in this paper)
154. Kruse, H; Grimme, S. (2012) J. Chem. Phys., 126, 154101.
HF-3c method:
155. Sure, R; Grimme, S. (2013) J. Comput. Chem., 34, 1672-1685 .
PBEh-3c method:
156. Grimme, S.; Brandenburg, J. G.; Bannwarth, C.; Hansen, A. (2015) Consistent structures and
interactions by density functional theory with small atomic orbital basis sets, J. Chem. Phys., 143,
054107 .
sTDA and sTD-DFT approaches for electronic spectra:
157. Grimme, S. (2013) J. Chem. Phys., 138, 244104 .
158. Risthaus, T.; Hansen, A.; Grimme, S. (2014) Phys. Chem. Chem. Phys., 16, 14408 .
159. Bannwarth, C.; Grimme, S. (2014) Comput. Theor. Chem., 1040-1041, 45-53 .
FOD analysis and FOD plots:

769
160. Grimme, S.; Hansen, A. (2015) A Practicable Real-Space Measure and Visualization of Static Electron-
Correlation Effects, Angew. Chem. Int. Ed., 54, 12308.
COSMO Implementation:
161. Sinnecker, S.; Rajendran, A.; Klamt, A.; Diedenhofen, M.; Neese, F. (2006) Calculation of Solvent Shifts
on Electronic G-Tensors with the Conductor-Like Screening Model (COSMO) and its Self-Consistent
Generalization to Real Solvents (COSMO-RS), J. Phys. Chem. A, 110, 22352245.
SOSCF Method:
162. Neese, F. (2000) Approximate Second Order Convergence for Spin Unrestricted Wavefunctions. Chem.
Phys. Lett., 325, 9398.
Relativity and SARC Basis Sets:
163. Pantazis, D. A.; Neese, F. (2012) All-Electron Scalar Relativistic Basis Sets for the 6p Elements,
Theor. Chem. Acc., 131, 1292.
164. Pantazis, D. A.; Neese, F. (2011) All-Electron Scalar Relativistic Basis Sets for the Actinides, J. Chem.
Theory Comput., 7, 677684.
165. Pantazis, D. A.; Neese, F. (2009) All-Electron Scalar Relativistic Basis Sets for the Lanthanides, J.
166. B
uhl, M.; Reimann, C.; Pantazis, D. A.; Bredow, T.; Neese, F. (2008) Geometries of Third-row
Transition-Metal Complexes from Density-Functional Theory J. Chem. Theory Comput., 4, 14491459.
167. Pantazis, D. A.; Chen, X.-Y.; Landis, C.R.; Neese, F. (2008) All Electron Scalar Relativistic Basis
Sets for Third Row Transition Metal Atoms. J. Chem. Theory Comput., 4, 908919.
QM/MM calculations with ORCA:
168. Sundararajan, M.; Neese, F. (2012) Detailed QM/MM study of the Electron Paramagnetic Resonance
Parameters of Nitrosyl Myoglobin, J. Chem. Theory Comput., 8, 563574.
169. Riplinger, C.; Neese, F. (2011) The reaction mechanism of Cytochrome P450 NO Reductase: A
Detailed Quantum Mechanics/Molecular Mechanics Study, ChemPhysChem, 12, 31923203.
170. Radoul, M.; Sundararajan, M.; Potapov, A.; Riplinger, C.; Neese, F.; Goldfarb, D. (2010) Revisiting
the nitrosyl complex of myoglobin by high-field pulse EPR spectroscopy and quantum mechanical
calculations, Phys. Chem. Chem. Phys., 12, 72767289.
171. Sundararajan, M.; Riplinger, C.; Orio, M.; Wennmohs, F.; Neese, F. (2009) Spectroscopic Properties
of Protein-Bound Cofactors: Calculation by Combined Quantum Mechanical/Molecular Mechanical
(QM/MM) Approaches, Encyc. Inorg. Chem., DOI: 10.1002/9781119951438.eibc0371.
172. Altun, A.; Kumar, D.; Neese, F.; Thiel, W. (2008) Multi-reference Ab Initio QM/MM Study on
Intermediates in the Catalytic Cycle of Cytochrome P450cam, J. Phys. Chem., 112, 1290412910.
Inorg. Chem., 45, 1105111059.
174. Sinnecker, S.; Neese, F. (2006) QM/MM Calculations with DFT for Taking into Account Protein
Effects on the EPR and Optical Spectra of Metalloproteins. Plastocyanin as a Case Study. J. Comp.
Chem. (Special issue on Theoretical Bioinorganic Chemistry), 27, 14631475.
175. Wanko, M.; Hoffmann, M.; Strodel, P.; Thiel, W.; Neese, F.; Frauenheim, T.; Elstner, M. (2005)
Calculating Absorption Shifts for Retinal Proteins: Computational Challenges J. Phys. Chem. B, 109,
36063615.
176. Schoneboom, J.; Neese, F.; Thiel, W. (2005) Towards Identification of the Compound I Reactive
Intermediate in Cytochrome P450 Chemistry: A QM/MM Study of its EPR and Mossbauer Parameters,
J. Am. Chem. Soc., 127, 58405853.
Applications that make use of ORCA include the following:
177. Kruse, H.; Mladek, A.; Gkionis, K.; Hansen, A.; Grimme, S.; Sponer J. (2015) Quantum Chemical
Benchmark Study on 46 RNA Backbone Families Using a Dinucleotide Unit, J. Chem. Theory Comput.,
11, 4972.
178. Qu, Z.-W.; Hansen, A.; Grimme, S. (2015) Co-C Bond Dissociation Energies in Cobalamin Derivatives
and Dispersion Effects: Anomaly or Just Challenging?, J. Chem. Theory Comput., 11, 1037.
179. A. Hansen, A.; Bannwarth, C.; Grimme, S.; Petrovic, P.; Werle, C.; Djukic, J.-P. (2014) The
Thermochemistry of London Dispersion-Driven Transition Metal Reactions: Getting the Right Answer
for the Right Reason, ChemistryOpen, 3, 177.
180. Krewald, V.; Neese, F.; Pantazis, D. A. (2013) On the magnetic and spectroscopic properties of
high-valent Mn3 CaO4 cubanes as structural units of natural and artificial water oxidizing catalysts, J.
Am. Chem. Soc., 135, 57265739.
181. Kampa, M.; Pandelia, M.-E.; Lubitz, W.; van Gastel, M.; Neese, F. (2013) A Metal-Metal Bond in
the Light-Induced State of [NiFe] Hydrogenases with Relevance to Hydrogen Evolution, J. Am. Chem.
Soc., 135, 39153925.
182. Pandelia, M.-E.; Bykov, D.; Izsak, R.; Infossi, P.; Giudici-Orticoni, M.-T.; Bill, E.; Neese, F.; Lubitz,
W. (2013) Electronic structure of the unique [4Fe-3S] cluster in O2 -tolerant hydrogenases characterized
by Fe-57 Mossbauer and EPR spectroscopy, Proc. Natl. Acad. Sci. USA, 110, 483488.
183. Atanasov, M.; Surawatanawong, P.; Wieghardt, K.; Neese, F. (2013) A theoretical study of zero-field
splitting in Fe(IV)S6 (S = 1) and Fe(III)S6 (S = 1/2) core complexes, [FeIV (Et2 dtc)3n (mnt)n ](n1)
and [FeIII (Et2 dtc)3n (mnt)n ]n (n = 0, 1, 2, 3): The origin of the magnetic anisotropy, Coord. Chem.
Rev., 257(1), 2741.
184. Retegan, M.; Collomb, M.-N.; Neese, F.; Duboc, C. (2013) A combined high-field EPR and quantum
chemical study on a weakly ferromagnetically coupled dinuclear Mn(III) complex. A complete analysis
of the EPR spectrum beyond the strong coupling limit, Phys. Chem. Chem. Phys., 15, 223234.
771
185. Zadrozny, J. M.; Atanasov, M.; Bryan, A. M.; Lin, C. Y.; Rekken, B. D.; Power, P. P.; Neese, F.; Long,
J. R. (2013) Slow magnetization dynamics in a series of two-coordinate iron(II) complexes, Chem.
Sci., 4, 125138.
186. Weber, K.; Kramer, T.; Shafaat, H. S.; Weyherm

uller, T.; Bill, E.; van Gastel, M.; Neese, F.; Lubitz,
W. (2012) A Functional [NiFe]-Hydrogenase Model Compound That Undergoes Biologically Relevant
Reversible Thiolate Protonation, J. Am. Chem. Soc., 134, 2074520755.
187. Kampa, M.; Lubitz, W.; van Gastel, M.; Neese, F.; (2012) Computational study of the electronic
structure and magnetic properties of the Ni-C state in [NiFe] hydrogenases including the second
coordination sphere, J. Biol. Inorg. Chem., 17, 12691281.
188. Atanasov, M.; Comba, P.; Helmle, S.; M

uller, D.; Neese, F. (2012) Zero-Field Splitting in a Series of
II
Structurally Related Mononuclear Ni -Bispidine Complexes, Inorg. Chem., 51, 1232412335.
189. Shafaat, H. S.; Weber, K.; Petrenko, T.; Neese, F.; Lubitz, W. (2012) Key Hydride Vibrational Modes
in [NiFe] Hydrogenase Model Compounds Studied by Resonance Raman Spectroscopy and Density
Functional Calculations, Inorg. Chem., 51, 1178711797.
190. Argirevic, T.; Riplinger, C.; Stubbe, J.; Neese, F.; Bennati, M. (2012) ENDOR Spectroscopy and
DFT Calculations: Evidence for the Hydrogen-Bond Network Within 2 in the PCET of E. coli
Ribonucleotide Reductase, J. Am. Chem. Soc., 134, 1766117670.
191. Albrecht, C.; Shi, L. L.; Perez, J. M.; van Gastel, M.; Schwieger, S.; Neese, F.; Streubel, R. (2012)
Deoxygenation of Coordinated Oxaphosphiranes: A New Route to P=C Double-Bond Systems, Chem.
Eur. J., 18, 97809783.
192. Maganas, D.; Krzystek, J.; Ferentinos, E.; Whyte, A. M.; Robertson, N.; Psycharis, V.; Terzis, A.;
Neese, F.; Kyritsis, P. (2012) Investigating Magnetostructural Correlations in the Pseudooctahedral
trans-[NiII {(OPPh2 ) (EPPh2 )N}2 (sol)2 ] Complexes (E = S, Se; sol = DMF, THF) by Magnetometry,
HFEPR, and ab Initio Quantum Chemistry, Inorg. Chem., 51, 72187231.
193. Ye, S. F.; Neese, F. (2012) How Do Heavier Halide Ligands Affect the Signs and Magnitudes of
the Zero-Field Splittings in Halogenonickel(II) Scorpionate Complexes? A Theoretical Investigation
Coupled to Ligand-Field Analysis, J. Chem. Theory Comput., 8, 23442351.
194. Nesterov, V.; Ozbolat-Schon, A.; Schnakenburg, G.; Shi, L. L.; Cangonul, A.; van Gastel, M.; Neese,
F.; Streubel, R. (2012) An Unusal Case of Facile Non-Degenerate P-C Bond Making and Breaking,
Chem. Asian J., 7, 17081712.
195. Bykov, D.; Neese, F. (2012) Reductive activation of the heme iron-nitrosyl intermediate in the reaction
mechanism of cytochrome c nitrite reductase: a theoretical study, J. Biol. Inorg. Chem., 17, 741760.
196. Lancaster, K. M.; Zaballa, M.E.; Sproules, S.; Sundararajan, M.; DeBeer, S.; Richards, J. H.; Vila, A.
J.; Neese, F.; Gray, H. B. (2012) Outer-Sphere Contributions to the Electronic Structure of Type Zero
Copper Proteins, J. Am. Chem. Soc., 134, 82418253.
197. Benkhauser-Schunk, C.; Wezisla, B.; Urbahn, K.; Kiehne, U.; Daniels, J.; Schnakenburg, G.; Neese, F.;
Lutzen, A. (2012) Synthesis, Chiral Resolution, and Absolute Configuration of Functionalized Trogers
Base Derivatives: Part II, ChemPlusChem, 77, 396403.
198. Ye, S. F.; Riplinger, C.; Hansen, A.; Krebs, C.; Bollinger, J. M.; Neese, F. (2012) Electronic
Structure Analysis of the Oxygen-Activation Mechanism by FeII - and -Ketoglutarate (KG)-Dependent
Dioxygenases, Chem. Eur. J., 18, 65556567.
199. Desrochers, P. J.; Sutton, C. A.; Abrams, M. L.; Ye, S. F.; Neese, F.; Telser, J.; Ozarowski, A.;
Krzystek, J. (2012) Electronic Structure of Nickel(II) and Zinc(II) Borohydrides from Spectroscopic
Measurements and Computational Modeling, Inorg. Chem., 51, 27932805.
200. Torres-Alacan, J.; Krahe, O.; Filippou, A. C.; Neese, F.; Schwarzer, D.; V
ohringer, P. (2012) The
III
Photochemistry of [Fe N3 (cyclam-ac)]PF6 at 266 nm, Chem. Eur. J., 18, 30433055.
201. Maekawa, M.; Romelt, M.; Daniliuc, C. G.; Jones, P. G.; White, P. S.; Neese, F.; Walter, M. D. (2012)
Reactivity studies on [Cp0 MnX(thf)]2 : manganese amide and polyhydride synthesis Chem. Sci., 3,
29722979.
202. Pantazis, D. A.; Ames, W.; Cox, N.; Lubitz, W.; Neese, F. (2012) Two interconvertible structures
that explain the spectroscopic properties of the oxygen-evolving complex of photosystem II in the S2
state, Angew. Chem. Int. Ed., 51, 99359940. (selected as cover article and VIP paper)
203. Vennekate, H.; Schwarzer, D.; Torres-Alacan, J.; Krahe, O.; Filippou, A. C.; Neese, F.; Vohringer, P.
(2012) Ultrafast primary processes of an iron-(III) azido complex in solution induced with 266 nm
light, Phys. Chem. Chem. Phys., 14, 61656172.
204. Christian, G. J.; Ye, S.; Neese, F. (2012) Oxygen activation in extradiol catecholate dioxygenases a
density functional study, Chem. Sci., 3, 16001611.
205. Cowley, R. E.; Christian, G. J.; Brennessel, W. W.; Neese, F.; Holland, P. L. (2012) A Reduced
(beta-Diketiminato)iron Complex with End-On and Side-On Nitriles: Strong Backbonding or Ligand
Non-Innocence? Eur. J. Inorg. Chem., 479483.
206. Thiessen, A.; Wettach, H.; Meerholz, K.; Neese, F.; Hoger, S.; Hertel, D. (2012) Control of electronic
properties of triphenylene by substitution, Organic Electronics, 13, 7183.
207. Lancaster, K. M.; Roemelt, M.; Ettenhuber, P.; Hu, Y. L.; Ribbe, M. W.; Neese, F.; Bergmann,
U.; DeBeer, S. (2011) X-ray Emission Spectroscopy Evidences a Central Carbon in the Nitrogenase
Iron-Molybdenum Cofactor, Science, 334, 974977.
208. Ames, W.; Pantazis, D. A.; Krewald, V.; Cox, N.; Messinger, J.; Lubitz, W.; Neese, F. (2011)
Theoretical Evaluation of Structural Models of the S2 State in the Oxygen Evolving Complex of
Photosystem II: Protonation States and Magnetic Interactions, J. Am. Chem. Soc., 133, 1974319757.
209. Antony, J.; Grimme, S.; Liakos, D. G.; Neese, F. (2011) Protein-Ligand Interaction Energies with
Dispersion Corrected Density Functional Theory and High-Level Wave Function Based Methods, J.
Phys. Chem. A, 115, 1121011220.
210. Radoul, M.; Bykov, D.; Rinaldo, S.; Cutruzzola, F.; Neese, F.; Goldfarb, D. (2011) Dynamic Hydrogen-
Bonding Network in the Distal Pocket of the Nitrosyl Complex of Pseudomonas aeruginosa cd(1)
Nitrite Reductase, J. Am. Chem. Soc., 133, 30433055.
773
211. Liakos, D. G.; Neese, F. (2011) Interplay of Correlation and Relativistic Effects in Correlated
Calculations on Transition-Metal Complexes: The Cu2 O2+
2 Core Revisited, J. Chem. Theory Comput.,
7, 15111523.
212. Riplinger, C.; Neese, F. (2011) The Reaction Mechanism of Cytochrome P450 NO Reductase: A
Detailed Quantum Mechanics/Molecular Mechanics Study, ChemPhysChem, 12, 31923203.
213. Maganas, D.; Sottini, S.; Kyritsis, P.; Groenen, E. J. J.; Neese, F. (2011) Theoretical Analysis of the
Spin Hamiltonian Parameters in CoII S4 Complexes, Using Density Functional Theory and Correlated
ab initio Methods, Inorg. Chem, 50, 87418754.
214. Surawatanawong, P.; Sproules, S.; Neese, F.; Wieghardt, K. (2011) Electronic Structures and Spec-
troscopy of the Electron Transfer Series Fe(NO)Lz2 (z = 1+, 0, 1, 2, 3; L = Dithiolene), Inorg.
Chem, 50, 1206412074.
215. Cox, N.; Ames, W.; Epel, B.; Kulik, L. V.; Rapatskiy, L.; Neese, F.; Messinger, J.; Wieghardt, K.;
Lubitz, W. (2011) Electronic Structure of a Weakly Antiferromagnetically Coupled Mn(II)Mn(III)
Model Relevant to Manganese Proteins: A Combined EPR, 55 Mn-ENDOR, and DFT Study, Inorg.
Chem, 50, 82388251.
216. Rota, J. B.; Knecht, S.; Fleig, T.; Ganyushin, D.; Saue, T.; Neese, F.; Bolvin, H. (2011) Zero field
splitting of the chalcogen diatomics using relativistic correlated wave-function methods, J. Chem.
Phys., 135, 114106.
217. Atanasov, M.; Ganyushin, D.; Pantazis, D. A.; Sivalingam, K.; Neese, F. (2011) Detailed Ab Initio
First-Principles Study of the Magnetic Anisotropy in a Family of Trigonal Pyramidal Iron(II) Pyrrolide
Complexes, Inorg. Chem, 50, 74607477.
218. Cox, N.; Rapatskiy, L.; Su, J. H.; Pantazis, D. A.; Sugiura, M.; Kulik, L.; Dorlet, P.; Rutherford, A.
W.; Neese, F.; Boussac, A.; Lubitz, W.; Messinger, J. (2011) Effect of Ca2+ /Sr2+ Substitution on the
Electronic Structure of the Oxygen-Evolving Complex of Photosystem II: A Combined Multifrequency
EPR, 55 Mn-ENDOR, and DFT Study of the S2 State, J. Am. Chem. Soc., 133, 36353648.
219. Maurice, R.; Sivalingam, K.; Ganyushin, D.; Guihery, N.; de Graaf, C.; Neese, F. (2011) Theoretical
Determination of the Zero-Field Splitting in Copper Acetate Monohydrate, Inorg. Chem, 50, 62296236.
220. Su, J. H.; Cox, N.; Ames, W.; Pantazis, D. A.; Rapatskiy, L.; Lohmiller, T.; Kulik, L. V.; Dorlet, P.;
Rutherford, A. W.; Neese, F.; Boussac, A.; Lubitz, W.; Messinger, J. (2011) The electronic structures
of the S2 states of the oxygen-evolving complexes of photosystem II in plants and cyanobacteria in the
presence and absence of methanol, Biochim. Biophys. Acta-Bioenergetics, 1807, 829840.
221. Bykov, D.; Neese, F. (2011) Substrate binding and activation in the active site of cytochrome c nitrite
reductase: a density functional study, J. Biol. Inorg. Chem, 16, 417430.
222. Gennari, M.; Orio, M.; Pecaut, J.; Bothe, E.; Neese, F.; Collomb, M. N.; Duboc, C. (2011) Influence
of Mixed Thiolate/Thioether versus Dithiolate Coordination on the Accessibility of the Uncommon
+I and +III Oxidation States for the Nickel Ion: An Experimental and Computational Study, Inorg.
Chem, 50, 37073716.
223. Ye, S. F.; Neese, F. (2011) Nonheme oxo-iron(IV) intermediates form an oxyl radical upon approaching
the C-H bond activation transition state, Proc. Natl. Acad. Sci. USA, 108, 12281233.
224. Gennari, M.; Pecaut, J.; DeBeer, S.; Neese, F.; Collomb, M. N.; Duboc, C. (2011) A Fully Delocalized
Mixed-Valence Bis--(Thiolato) Dicopper Complex: A Structural and Functional Model of the Biological
Cu(A) Center, Angew. Chem., Int. Ed., 50, 56615665.
225. Gennari, M.; Retegan, M.; DeBeer, S.; Pecaut, J.; Neese, F.; Collomb, M. N.; Duboc, C. (2011)
Experimental and Computational Investigation of Thiolate Alkylation in Ni(II) and Zn(II) Complexes:
Role of the Metal on the Sulfur Nucleophilicity, Inorg. Chem, 50, 1004710055.
226. Atanasov, M.; Delley, B.; Neese, F.; Tregenna-Piggott, P. L.; Sigrist, M. (2011) Theoretical Insights
into the Magnetostructural Correlations in Mn(3)-Based Single-Molecule Magnets, Inorg. Chem, 50,
21122124.
227. Neese, F.; Pantazis, D. A. (2011) What is not required to make a single molecule magnet, Faraday
Discussions, 148, 229238.
228. Lassalle-Kaiser, B.; Hureau, C.; Pantazis, D. A.; Pushkar, Y.; Guillot, R.; Yachandra, V. K.; Yano, J.;
Neese, F.; Anxolabeh`ere-Mallart, E. (2010) Activation of a water molecule using a mononuclear Mn
complex: from Mn-aquo, to Mn-hydroxo, to Mn-oxyl via charge compensation, Energy Environ. Sci.,
3, 924938.
229. Pantazis, D. A.; Krewald, V.; Orio, M.; Neese, F. (2010) Theoretical magnetochemistry of dinuclear
manganese complexes: broken symmetry density functional theory investigation on the influence of
bridging motifs on structure and magnetism, Dalton Trans., 39, 49594967.
230. Woertink, J. S.; Tian, L.; Maiti, D.; Lucas, H. R.; Himes, R. A.; Karlin, K. D.; Neese, F.; Wurtele,
C.; Holthausen, M. C.; Bill, E.; Sundermeyer, J.; Schindler, S.; Solomon, E. I. (2010) Spectroscopic
and Computational Studies of an End-on Bound Superoxo-Cu(II) Complex: Geometric and Electronic
Factors That Determine the Ground State, Inorg. Chem, 49, 94509459.
231. McNaughton, R. L.; Roemelt, M.; Chin, J. M.; Schrock, R. R.; Neese, F.; Hoffman, B. M. (2010)
Experimental and Theoretical EPR Study of JahnTeller-Active HIPTN(3)N MoL Complexes (L =
N2 , CO, NH3 ), J. Am. Chem. Soc., 132, 86458656.
232. Geng, C. Y.; Ye, S. F.; Neese, F. (2010) Analysis of Reaction Channels for Alkane Hydroxylation by
Nonheme Iron(IV)-Oxo Complexes, Angew. Chem., Int. Ed., 49, 57175720.
233. Orio, M.; Jarjayes, O.; Kanso, H.; Philouze, C.; Neese, F.; Thomas, F. (2010) X-Ray Structures
of Copper(II) and Nickel(II) Radical Salen Complexes: The Preference of Galactose Oxidase for
Copper(II), Angew. Chem., Int. Ed., 49, 49894992.
234. Gennari, M.; Orio, M.; Pecaut, J.; Neese, F.; Collomb, M. N.; Duboc, C. (2010) Reversible Apical
Coordination of Imidazole between the Ni(III) and Ni(II) Oxidation States of a Dithiolate Complex: A
Process Related to the Ni Superoxide Dismutase, Inorg. Chem, 49, 63996401.
235. Ye, S. F.; Price, J. C.; Barr, E. W.; Green, M. T.; Bollinger, J. M.; Krebs, C.; Neese, F. (2010)
Cryoreduction of the NO-Adduct of Taurine:alpha-Ketoglutarate Dioxygenase (TauD) Yields an Elusive
{FeNO}8 Species, J. Am. Chem. Soc., 132, 47394751.
775
236. Maganas, D.; Grigoropoulos, A.; Staniland, S. S.; Chatziefthimiou, S. D.; Harrison, A.; Robertson, N.;
Kyritsis, P.; Neese, F. (2010) Tetrahedral and Square Planar Ni(SPR2 )2 N2 complexes, R = Ph & iPr
Revisited: Experimental and Theoretical Analysis of Interconversion Pathways, Structural Preferences,
and Spin Delocalization, Inorg. Chem, 49, 50795093.
237. Anoop, A.; Thiel, W.; Neese, F. (2010) A Local Pair Natural Orbital Coupled Cluster Study of Rh
Catalyzed Asymmetric Olefin Hydrogenation, J. Chem. Theory Comput., 6, 31373144.
238. Duboc, C.; Collomb, M. N.; Neese, F. (2010) Understanding the Zero-Field Splitting of Mononuclear
Manganese(II) Complexes from Combined EPR Spectroscopy and Quantum Chemistry, Appl. Magn.
Res., 37, 229245.
239. Ye, S. F.; Neese, F. (2010) The Unusual Electronic Structure of Dinitrosyl Iron Complexes, J. Am.
Chem. Soc., 132, 36463647.
240. Kochem, A.; Orio, M.; Jarjayes, O.; Neese, F.; Thomas, F. (2010) Unsymmetrical one-electron oxidized
Ni(II)-bis(salicylidene) complexes: a protonation-induced shift of the oxidation site, Chem. Commun.,
46, 67656767.
241. Ozbolat-Schon, A.; Bode, M.; Schnakenburg, G.; Anoop, A.; van Gastel, M.; Neese, F.; Streubel, R.
(2010) Insights into the Chemistry of Transient P-Chlorophosphanyl Complexes, Angew. Chem., Int.
Ed., 49, 68946898.
242. Vancoillie, S.; Chalupsky, J.; Ryde, U.; Solomon, E. I.; Pierloot, K.; Neese, F.; Rulisek, L. (2010)
Multireference Ab Initio Calculations of g tensors for Trinuclear Copper Clusters in Multicopper
Oxidases, J. Phys. Chem. B, 114, 76927702.
243. Grote, D.; Finke, C.; Kossmann, S.; Neese, F.; Sander, W. (2010) 3,4,5,6-Tetrafluorophenylnitren-2-yl:
A Ground-State Quartet Triradical, Chem. Eur. J., 16, 44964506.
244. Ye, S. F.; Neese, F.; Ozarowski, A.; Smirnov, D.; Krzystek, J.; Telser, J.; Liao, J. H.; Hung, C. H.;
Chu, W. C.; Tsai, Y. F.; Wang, R. C.; Chen, K. Y.; Hsu, H. F. (2010) Family of V(III)-Tristhiolato
Complexes Relevant to Functional Models of Vanadium Nitrogenase: Synthesis and Electronic Structure
Investigations by Means of High-Frequency and -Field Electron Paramagnetic Resonance Coupled to
Quantum Chemical Computations, Inorg. Chem, 49, 977988.
245. Hegele, P.; Santhamma, B.; Schnakenburg, G.; Frohlich, R.; Kataeva, O.; Nieger, M.; Kotsis, K.;
Neese, F.; Dotz, K. H. (2010) Hydroquinoid Chromium Complexes Bearing an Acyclic Conjugated
Bridge: Chromium-Templated Synthesis, Molecular Structure, and Haptotropic Metal Migration,
Organometallics, 29, 61726185.
246. Ye, S. F.; Neese, F. (2010) Accurate Modeling of Spin-State Energetics in Spin-Crossover Systems
with Modern Density Functional Theory, Inorg. Chem, 49, 772774.
247. Orio, M.; Philouze, C.; Jarjayes, O.; Neese, F.; Thomas, F. (2010) Spin Interaction in Octahedral Zinc
Complexes of Mono- and Diradical Schiff and Mannich Bases, Inorg. Chem, 49, 646658.
248. Pantazis, D. A.; Orio, M.; Petrenko, T.; Zein, S.; Lubitz, W.; Messinger, J.; Neese, F. (2009) Structure
of the Oxygen-Evolving Complex of Photosystem II: Information on the S2 state through Quantum
Chemical Calculation of its Magnetic Properties. Phys. Chem. Chem. Phys., 11, 67886798.
249. Baffert, C.; Orio, M.; Pantazis, D. A.; Duboc, C.; Blackman, A.G.; Blondin, G.; Neese, F.; Deronzier,A.;
Collomb, M-N. (2009) A trinuclear terpyridine frustrated spin system with a MnIV 3 O4 core: synthesis,
physical characterization and quantum chemical modeling of its magnetic properties. Inorg. Chem.,
48, 1028110288.
250. Liakos, D.; Neese, F. (2009) A multiconfigurational ab initio study of the zero-field splitting in the di-
and trivalent hexaquo-chromium complexes. Inorg. Chem., 48, 1057210580.
251. Astashkin, A.V.; Klein, E.C.; Ganyushin, D.; Johnson.Winters, K.; Neese, F.; Kappler, U.; Enemark,
J.H. (2009) Exchangeable oxygens in the vicinity of the molybdenum center of the high-pH form of
sulfite oxidase and sulfite dehydrogenase. Phys. Chem. Chem. Phys., 11, 67336742.
252. Orio, M.; Pantazis, D. A.; Petrenko, T.; Neese, F. (2009) Magnetic and spectroscopic properties of
mixed valence manganese(III,IV) dimers: a systematic study using broken symmetry density functional
theory, Inorg. Chem., 48, 72517260.
253. Klein, E.L.; Astashkin, A.V.; Ganyushin, D.; Johnson-Winters, K.; Wilson, H.L.; Rajagopalan, K.
V.; Neese, F.; Enemark, J.H. (2009) Direct Detection and Characterization of Chloride in the Active
Site of the Low-pH Form of Sulfite Oxidase Using ESEEM Spectroscopy, Isotopic Labeling, and DFT
Calculations, Inorg. Chem., 48(11), 47434752.
254. Vancoillie, S.; Rulisek, L.; Neese, F.; Pierloot, K. (2009) Theoretical description of the structure and
magnetic properties of nitroxide-Cu(II)-nitroxide spin triads, J. Phys. Chem., 113, 61496157.
255. Cowley, R.E.; Bill, E.; Neese, F.; Brennessel,W.W.; Holland, P.L. (2009) Iron(II) Complexes With
Redox-Active Tetrazene (RNNNNR) Ligands, Inorg. Chem., 48, 48284836.
256. Gansauer, A.; Fleckhaus, A.; Lafon, A.; Okkel, M.; Anakuthil, A.; Kotsis, K.; Neese, F. (2009)
Catalysis via Homolytic Substitutions with C-O and Ti-O Bonds: Oxidative Additions and Reductive
Eliminations in Single Electron Steps. J. Am. Chem. Soc., 131, 1698916999.
257. Ye, S.; Neese, F. (2009) Quantum Chemical Studies of C-H Activation Reactions by High-Valent
Nonheme Iron Centers Curr. Op. Chem. Biol., 13(1), 8998.
258. Krahe, O.; Neese, F.; Streubel, R. (2009) The quest for ring-opening of oxaphosphirane complexes: a
coupled cluster and density functional study of CH3 PO isomers and their Cr(CO)5 complexes Chem.
Eur. J., 15, 25942601.
259. Romain, S.; Duboc, C.; Neese, F.; Riviere, E.; Hanton, L. R.; Blackman, A. G.; Philouze, C.; Lepretre,
J. C.; Deronzier, A.; Collomb, M. N. (2009) An Unusual Stable Mononuclear Mn(III) Bis-terpyridine
Complex Exhibiting Jahn-Teller Compression: Electrochemical Synthesis, Physical Characterisation
and Theoretical Study, Chem. Eur. J., 15, 980988
260. Zein, S.; Neese, F. (2008) Ab initio and Coupled Perturbed DFT Calculation of Zero-Field Splittings
in Mn(II) Transition Metal complexes. J. Phys. Chem. A, 112, 79767983.
261. Ye, S.; Tuttle, T.; Bill, E.; Gross, Z.; Thiel, W.; Neese, F. (2008) The Noninnocence of Iron Corroles:
A combined Experimental and Quantum Chemical Study. Chem. Eur. J. (selected as very important
paper), 34, 1083910851.
777
262. Duboc, C.; Collomb, M.-N.; Pecaut, J.; Deronzier, A.; Neese, F. (2008) Definition of Magneto-Structural
Correlations for the Mn(II) Ion. Chem. Eur. J., 21, 64986509.
263. Berry, J.F.; DeBeer-George, S.; Neese, F. (2008) Electronic Structure and Spectroscopy of Superoxi-
dized Iron Centers in Model Systems: Theoretical and Experimental Trends. Phys. Chem. Chem. Phys.,
10, 43614374.
264. Sander, W.; Grote, D.; Kossmann, S.; Neese, F. (2008) 2.3.5.6-Tetrafluorophenylnitren-4-yl: EPR
Spectroscopic Characterization of a Quartet Ground State Nitreno Radical, J. Am. Chem. Soc., 130,
43964403.
265. Scheifele, Q.; Riplinger, C.; Neese, F.; Weihe, H.; Barra, A.L.; Jurany, F.; Podlesnyak, A.; Tregenna-
Piggot, P.W.L. (2008) Spectroscopic and Theoretical Study of a Mononuclear Mn(III) Bioinorganic
Complex Exhibiting a Compressed Jahn-Teller Octahedron, Inorg. Chem., 47, 439447.
266. Zein, S.; Kulik, L.V.; Yano, J.; Kern, J.; Zouni, A.; Yachandra, V.K.; Lubitz, W.; Neese, F.; Messinger,
J. (2008) Focussing the View on Natures Water Splitting Catalyst Phil. Trans. Roy. Soc. London B,
363, 11671177.
267. Zein, S.; Duboc, C.; Lubitz, W.; Neese, F. (2008) Theoretical Characterization of zero-Field Splittings
in Mn(II) Complexes. Inorg. Chem., 47, 134142.
268. Parker, D.J.; Hammond, D.; Davies, E.S.; Garner, C.D.; Benisvy, L.; McMaster, J.; Wilson, C.; Neese,
F.; Bothe, E.; Bittl, R.; Teutloff, C. (2007) A stable H-bonded ortho-Thioether Phenoxyl-Radical: A
Chemical and Spectroscopic Analogue of Tyr272 in apo-Galactose Oxidase, J. Biol. Inorg. Chem. (Ed
Stiefel memorial issue), 101, 18591864.
269. Chlopek, K.; Muresan, N.; Neese, F.; Wieghardt, K. (2007) Electronic Structures of Five-Coordinate
Complexes of Iron Containing Zero, One, or Two Radical Ligands: A Broken Symmetry Density
Functional Theoretical Study, Chem. Eur. J., 13, 83918403.
270. Muresan, N.; Chlopek, K.; Weyherm uller, T.; Neese, F.; Wieghardt, K. (2007) Bis(-diimine)nickel
Complexes: Molecular and Electronic Structure of Three Members of the Electron-Transfer Series
[Ni(L)2 ]z (z = 0, 1+, 2+) (L = 2-Phenyl-1,4-bis(isopropyl)-1,4-diazabutadiene). A Combined Experi-
mental and Theoretical Study, Inorg. Chem., 46, 49054916.
271. Ray, K.; Petrenko, T.; Wieghardt, K.; Neese, F. (2007) Joint Spectroscopic and Theoretical Investiga-
tions of Transition Metal Complexes Involving Non-Innocent Ligands. Dalton Trans., 1552 (selected
for cover picture).
272. Sinnecker, S.; Svensen, N.; Barr, E.W.; Ye, S.; Bollinger, J.M.; Neese, F.; Krebs, C. (2007) Spectroscopic
and Theoretical Evaluation of the Structure of the High-Spin Fe(IV)-Oxo Intermediates in Taurine:-
Ketoglutarate Dioxygenase from Escherichia coli and its His99Ala Ligand Variant J. Am. Chem. Soc.,
129, 61686179.
273. Duboc, C.; Phoeung, T; Zein, S.; Pecaut, J.; Collomb, M.-N.; Neese. F. (2007) Origin of the zero field
splitting in mononuclear dihalide Mn(II) complexes: an investigation by multifrequency high-field EPR
and density functional theory (DFT), Inorg. Chem., 46, 49054916.
274. DeBeer-George, S.; Petrenko, T.; Aliaga-Alcade, N.; Bill, E.; Mienert, B.; Sturhan, W.; Ming, Y.;
Wieghardt, K.; Neese, F. (2007) Characterization of a Genuine Iron(V)Nitrido Species by Nuclear
Resonant Vibrational Spectroscopy Coupled to Density Functional Calculations, J. Am. Chem. Soc.,
129, 1105311060.
275. Lehnert, N.M; Cornelissen, U.; Neese, F.; Ono, T.; Noguchi, Y.; Okamoto, K.-I.; Fujisawa, K. (2007)
Synthesis and Spectroscopic Characterization of Cu(II)-Nitrite Complexes with Hydrotris(pyrazolyl)borate
and Related Ligands. Inorg. Chem., 46, 39163933.
276. Carmieli, R.; Larsen, T.; Reed, G.H.; Zein, S.; Neese, F.; Goldfarb, D. (2007) The Catalytic Mn2+
Sites in the Enolase-Inhibitor Complex - Crystallography, Single Crystal EPR and DFT calculations.
J. Am. Chem. Soc., 129, 42404252.
277. Kokatam, S.; Ray, K.; Pap, J.; Bill, E.; Geiger, W.E.; LeSuer, R.J.; Rieger, P.H.; Weyherm
uller, T.;
Neese, F.; Wieghardt, K. (2007) Molecular and Electronic Structure of Square Planar Gold Complexes
Containing Two 1,2-di(4-tert-butylphenyl)ethylene-1,2-dithiolato Ligands: [Au(L)2 ]1+/0/1/2 . A
Combined Experimental and Computational Study, Inorg. Chem., 46, 11001111.
278. Ray, K.; DeBeer-George, S.; Solomon, E.I.; Wieghardt, K.; Neese, F. (2007) Description of the Ground
State Covalencies of the Bis(dithiolato)Transition Metal Complexes Using X-ray Absorption Spectral
and Time-Dependent-Density-Functional Studies. Chem. Eur. Journal, 13(10), 2753 (selected for
cover picture).
Inorg. Chem., 45, 1105111059.
280. Bart, S.C.; Chlopek, K.; Bill, E., Bouwkamp, B.W.; Lobkovsky, E.; Neese, F.; Wieghardt, K.; Chirik,
P.J. (2006) Electronic Structure of Bis(imino)pyridine Iron Dichloride, Monochloride and Neutral
Ligand Complexes: A Combined Structural, Spectroscopic and Computational Study, J. Am. Chem.
Soc., 128, 1390113912.
281. Patra, A.K.; Bill, E.; Bothe, E.; Chlopek, K.; Neese, F.; Weyherm
uller, T.; Stobie, K.; Ward, M.D.;
McCleverty, J.A.; Wieghardt, K. (2006) The Electronic Structure of Mononuclear Bis(1,2-diaryl-1,2-
ethylenedithiolate)iron Complexes Containing a Fifth Cyanide or Phosphite Ligand: A Combined
Experimental and Computational Study, Inorg. Chem., 45, 78777890.
282. Berry, J.F.; Bill, E.; Bothe, E.; DeBeer-George, S.; Mienert, B.; Neese, F.; Wieghardt, K. (2006) An
Octahedral Coordination Complex of Iron(VI) One Step Ahead of Nature?, Science, 312, 19371941.
283. Petrenko, T.; Ray, K.; Wieghardt, K.; Neese, F. (2006) Vibrational Markers for the Open-Shell
Character of Metal bis-Dithiolenes: An Infrared, resonance Raman and Quantum Chemical Study. J.
Am. Chem. Soc., 128, 44224436.
284. Chlopek, K.; Bothe, E.; Neese, F.; Weyherm uller, T.; Wieghardt, K. (2006) The Molecular and
Electronic Structures of Tetrahedral Complexes of Nickel and Cobalt Containing N, N 0 -Disubstituted,
Bulky o-Diiminobenzosemiquinonate(1-) -Radical Ligands, Inorg. Chem., 45, 62986307.
779
285. Kababya, S.; Nelson. J.; Calle, C.; Neese, F.; Goldfarb, D. (2006) The electronic structure of bi-nuclear
mixed valent copper azacryptates derived from integrated advanced EPR and DFT calculations. J.
Am. Chem. Soc., 128, 20172029.
286. Berry, J.F.; Bill, E.; Neese, F.; Garcia-Serres, R.; Weyherm
uller, T.; Wieghardt, K. (2006) Effect
of N-Methylation of Macrocyclic Amine Ligands on the Spin State of Fe(III): A Tale of Two Fluoro
Complexes. Inorg. Chem., 45, 20272037.
287. Kapre, R.; Ray, K.; Sylvestre, I.; Weyherm

uller, T.; DeBeer-George, S.; Neese, F.; Wieghardt, K. (2006)
The Molecular and Electronic Structure of Oxo-bis(benzene-1,2-dithiolato)chromate(V) Monoanions.
A Combined Experimental and Density Functional Study. Inorg. Chem., 45, 34993509.
288. Zhu, W.; Marr, A.C.; Wang, Q.; Neese, F.; Spencer, J.E.; Blake, A.J.; Cooke, P.A.; Wilson, C.;
Schr
oder, M. (2005) Modulation of the Electronic Structure and the Ni-Fe Distance in Heterobimetallic
Models for the Active Site in [NiFe]Hydrogenase: Is there a Ni-Fe Bond? Proc. Natl. Acad. Sci.
(USA), 102, 1828018285.
289. Astashkin, A.V.; Neese, F.; Raitsimaring, A.M.; Cooney, J.J.A.; Bultman, E.; Enemark, J.H. (2005)
Pulsed EPR investigation of systems modelling molybdenum enzymes: hyperfine and quadrupole
parameters of oxo-17 O in [Mo17 O(SPh)4 ] , J. Am. Chem. Soc., 127, 1671316722.
290. Benisvy, L.; Bittl, R.; Bothe, E.; Garner, C.D.; McMaster, J.; Ross, S.; Teutloff, C.; Neese, F. (2005)
Phenoxyl Radicals Hydrogen-Bonded to Imidazolium Analogues of Tyrosyl D of Photosystem II:
High-Field EPR and DFT Studies. Angew. Chem. Int. Ed., 44, 53145317.
291. Praneeth, V.K.K.; Neese, F.; Lehnert, N. (2005) Spin Density Distribution in Five- and Six-Coordinate
Iron(II)-Porphyrin NO Complexes Evidenced by Magnetic Circular Dichroism Spectroscopy. Inorg.
Chem., 44, 25702572.
292. Sinnecker, S.; Neese, F.; Lubitz, W. (2005) Dimanganese Catalase Spectroscopic Parameters from
Broken Symmetry Density Functional Theory of the Superoxidized MnIII /MnIV state, J. Biol. Inorg.
Chem., 10, 231238.
293. Blanchard, S.; Neese, F.; Bothe, E.; Bill, E.; Weyherm
uller, T.; Wieghardt, K. (2005) Square Planar
vs. Tetrahedral Coordination in Diamagnetic Complexes of Nickel(II) Containing Two Bidentate
Radical Monoanions, Inorg. Chem., 44, 36363656.
294. Mader-Cosper, M.; Neese, F.; Astashkin, A.V.; Carducci, M.A.; Raitsimring, A.M.; Enemark, J.H.
(2005) Determination of the Magnitude and Orientation of the g-Tensors for cis,trans-(L-N2 S2 )MoV OX
(X=Cl, SCH2 Ph) by Single Crystal EPR and Molecular Orbital Calculations, Inorg. Chem., 44,
12901301.
295. Fouqeau, A.; Casida, M.E.; Lawson, L.M.; Hauser, A.; Neese, F. (2005) Comparison of Density
Functionals for Energy and Structural Differences Between the High-[5 T2g : (t42g )(e2g )] and Low-[1 A1g :
(t62g )(e0g )] Spin States of Iron(II) Coordination Compounds: II. Comparison of Results for More than
Ten Modern Functionals with Ligand Field Theory and Ab Initio Results for Hexaquoferrous Dication,
[Fe(H2 O)6 ]2+ and Hexaminoferrous Dication [Fe(NH3 )6 ]2+ , J. Chem. Phys., 122, 044110.
296. Aliaga-Alcade, N.; DeBeer George, S.; Bill, E.; Wieghardt, K.; Neese, F. (2005) The Geometric and
Electronic Structure of [(Cyclam-acetato)Fe(N)]+: a Genuine Iron(V) Species with Ground State Spin
S = 1/2. Angew. Chem. Int. Ed., 44, 29082912.
297. Bill, E.; Bothe, E.; Chaudhuri, P.; Chlopek, K.; Herebian, D.; Kokatam, S.; Ray, K. Weyherm
uller, T.;
Neese, F.; Wieghardt, K. (2004) Molecular and Electronic Structure of Four- and Five-Coordinate
Cobalt Complexes Containing Two o-Phenylendiamine- or Two o-Aminophenol-Type Ligands at Various
Oxidation Levels: An Experimental, Density Functional and Correlated ab initio Study. Chem. Eur.
J., 11, 204224.
298. Paine, T.; Bothe, W.; Bill, E.; Weyherm

uller, T.; Slep, L.; Neese, F.; Chaudhuri, P. (2004) Nonoxo
Vanadium(IV) and Vanadyl(V) Complexes with Mixed O,X,O-Donor Ligand (X = S, Se, P, PO), Inorg.
Chem., 43, 73247338.
299. Baute, D.; Arieli, D.; Zimmermann, H.; Neese, F.; Weckhuysen, B.; Goldfarb, D. (2004) The Structure
of Copper Histidine Complexes in Solution and in Zeolite Y: A Combined X- and W-Band Pulsed
EPR/ENDOR and DFT Study, J. Am. Chem. Soc., 126, 1173311745.
300. Garcia Serres R.; Grapperhaus, C.A.; Bothe, E.; Bill, E.; Weyherm
uller, T.; Neese, F.; Wieghardt, K.
(2004) Structural, Spectroscopic and Computational Study of an Octahedral, Non-heme {FeNO}6,7,8
Series: [Fe(NO)(cyclam-ac)]2+/1+/0 , J. Am. Chem. Soc., 126, 51385153.
301. Sinnecker, S.; Noodleman, L.; Neese, F.; Lubitz, W. (2004) Calculation of the EPR Parameters of a
Mixed Valence Mn(III)/Mn(IV) Model Complex with Broken Symmetry Density Functional Theory. J.
Am. Chem. Soc., 126, 26132622.
302. Sinnecker, S.; Neese, F.; Lubitz, W. (2004) Benzosemichinone Solvent Interactions. A Density
Functional Study of Electric and Magnetic Properties for Probing Hydrogen Bond Strengths and
Geometries. J. Am. Chem. Soc., 126, 32803290.
303. van Gastel, M.; Fichtner, C.; Neese, F.; Lubitz, W. (2005) EPR Experiments to Elucidate the Structure
of the Ready and Unready States of the [NiFe] Hydrogenase of Desulfovibrio vulgaris Miyazaki F.
Biochem. Soc. Trans., 33, 711.
304. van Gastel, M.; Lassman, G.; Lubitz, W.; Neese, F. (2004) The unusual EPR parameters of the
cysteine radical: a DFT and correlated ab initio study J. Am. Chem. Soc., 126, 22372246.
305. Fouqueau, A.; Mer, S.; Casida, M.E.; Daku, L.M.L.; Hauser, A.; Mieva, T.; Neese, F. (2004)
Comparison of Density Functionals for Energy and Structural Differences between the High [5 T2g :
t42g e2g ] and Low [1 A1g : t62g ] Spin States of the Hexaquo-Ferrous Ion, [Fe(H2 O)6 ]2+ , J. Chem. Phys.,
120, 94739486.
306. Slep, L.D.; Mijovilovich, A.; Meyer-Klaucke, W.; Weyherm

uller, T.; Bill, E.; Bothe, E.; Neese, F.;
Wieghardt, K. (2003) The Mixed-valent {Fe (-O)(-carboxylato)2 FeIII }3+ Core. J. Am. Chem.
IV
Soc., 125, 1555415570.
307. Herebian, D.; Wieghardt, K.; Neese, F. (2003) Analysis and Interpretation of Metal-Radical Coupling in
a Series of Square Planar Nickel Complexes. Correlated Ab Initio and Density Functional Investigation
of [Ni(LISQ )2 ] (LISQ =3,5-di-tert-butyl-odiiminobenzosemquinone). J. Am. Chem. Soc., 125, 10997
11005.
781
308. Herebian, D.; Bothe, E.; Neese, F.; Weyherm uller, T.; Wieghardt, K. (2003) The Molecular and
Electronic Structures of Bis(o-diiminobenzosemiquinonato)metal(II) Complexes (Ni, Pd, Pt), their
Monocations and Anions, and their Dimeric Dications Containing Weak Metal-Metal Bonds. J. Am.
Chem. Soc., 125, 91169128.
309. Ghosh, P.; Bill, E.; Weyherm uller, T.; Neese, F.; Wieghardt, K. (2003) The non-Innocence of the
Ligand Glyoxal-bis (2-mercaptoanil). The Electronic Structures of [Fe(gma)]2 , [Fe(gma)(py)] py,
[Fe(gma)(CN)]1/0 , [Fe(gma)I], [Fe(gma)(PR3 )n ] (n = 1, 2). Experimental and Theoretical Evidence
for Excited State Coordination. J. Am. Chem. Soc., 125, 12931308.
310. Einsle, O.; Messerschmidt, A.; Huber, R.; Kroneck, P.M.H.; Neese, F. (2002) Mechanism of the Six
Electron Reduction of Nitrite to Ammonia by Cytochrome c Nitrite Reductase (CCNIR). J. Am. Chem.
Soc., 124, 1173711745.
311. Sun, X.; Chun, H.; Hildenbrand, K.; Bothe, E.; Weyherm uller, T.; Neese, F.; Wieghardt, K.
(2002) o-Iminobenzosemiquinonato(1-) and o-Amidophenolato(2-) Complexes of Palladium(II) and
Plantinum(II): A Combined Experimental and Density Functional Theoretical Study, Inorg. Chem.,
41, 42954303.
312. Li, M.; Bonnet, D.; Bill, E.; Neese, F.; Weyherm
uller, T.; Blum, N.; Sellmann, D.; Wieghardt, K. (2002)
Tuning the Electronic Structure of Octahedral Iron Complexes [FeL(X)] (L = 1-alkyl-4,7-bis(4-tert-
butyl-2-mercaptobenzyl)-1,4,7-triazacyclo-nonane, X = Cl, CH3 O, CN, CO). The S = 1/2 S = 3/2
Spin-Equilibrium of [FeLP r (NO)]. Inorg. Chem., 41, 34443456.
313. Lehnert, N.; Neese, F.; Ho, R.Y.N.; Que Jr., L.; Solomon, E.I. (2002) Electronic Structure and
Reactivity of Low-Spin Fe(III)-Hydroperoxo Complexes: Comparison to Activated Bleomycin. J. Am.
Chem. Soc., 124, 1081010822.
314. Grapperhaus, C.A.; Bill, E.; Weyherm uller, T.; Neese, F.; Wieghardt, K. (2001) Electronic and Geo-
metric Structure and Spectroscopy of a High Valent Manganese(V) Nitrido Complex. An Experimental
and DFT Study. Inorg. Chem., 41, 41914198.
315. Neese, F., Solomon, E.I. (1998) Detailed Spectroscopic and Theoretical Studies on [Fe(EDTA)(O2 )]3 :
the Electronic Structure of the Side-On Ferric Peroxide Bond and its Relevance to Reactivity. J. Am.
Chem. Soc., 120, 1282912848.
Some Reviews of interest:
316. Neese, F.; Liakos, D. G.; Ye, S. F. (2011) Correlated Wavefunction Methods in Bioinorganic Chemistry,
J. Biol. Inorg. Chem, 16, 821829.
317. Neese, F.; Ames, W.; Christian, G.; Kampa, M.; Liakos, D. G.; Pantazis, D. A.; Roemelt, M.;
Surawatanawong, P.; Ye, S. F. (2010) Dealing with Complexity in Open-Shell Transition Metal
Chemistry from a Theoretical Perspective: Reaction Pathways, Bonding, Spectroscopy, and Magnetic
Properties, Adv. Inorg. Chem., 62, 301349.
318. Orio, M.; Pantazis, D. A.; Neese, F. (2009) Density Functional Theory, Photosynth. Res., 102, 443453.
319. Neese, F. (2009), Density Functional Theory and EPR Spectroscopy: a guided tour. EPR Newsletter,
18(4), Pro & Contra section.
320. Neese, F. (2009) Prediction of Molecular Spectra and Molecular Properties with Density Functional
Theory: from Fundamental Theory to Exchange Coupling. Coord. Chem. Rev., 253, 526563.
321. Neese, F. (2009) Spin Hamiltonian Parameters from First Principle Calculations: Theory and Applica-
tion. In: Hanseon, G.; Berliner, L. (Eds.) Biological Magnetic Resonance. Vol 28, pp 175232.
322. Ray, K.; Petrenko, T.; Wieghardt, K.; Neese, F. (2007) Joint Spectroscopic and Theoretical Investiga-
tions of Transition Metal Complexes Involving Non-Innocent Ligands. Dalton Trans., 1552. (selected
for cover picture)
323. Kirchner, B.; Wennmohs, F.; Ye, S.; Neese, F. (2007) Theoretical Bioinorganic Chemistry: Electronic
Structure Makes a Difference, Curr. Op. Chem. Biol., 11, 131141.
324. Neese, F.; Petrenko, T.; Ganyushin, D.; Olbrich, G. (2007) Advanced Aspects of ab initio Theoretical
Spectroscopy of Open-Shell Transition Metal Ions. Coord. Chem. Rev., 205, 288327.
325. Ye, S.; Neese, F. (2006) Combined Quantum Chemical and Spectroscopic Studies on Transition Metal
Complexes with Coordinating Radicals. Chemtracts (Special Volume on Computational Inorganic
Chemistry), 19, 7786.
326. Sinnecker, S.; Neese, F. (2006) Theoretical Bioinorganic Spectroscopy, Invited Chapter in the Series
Current Topics in Chemistry, Editor M. Reiher, Springer, Heidelberg.
327. Neese, F. (2006) Quantum Chemical Approaches to Spin-Hamiltonian Parameters. Specialist Periodical
Reports on EPR Spectroscopy Vol. 20, (Ed. B. Gilbert) Royal Scoiety Press.
328. Neese, F. (2006) A Critical Evaluation of DFT, including Time-Dependent DFT, Applied to Bioinor-
ganic Chemistry. J. Biol. Inorg. Chem., (commentary on invitation), 11, 702711.
329. Neese, F.; Munzarova, M.L. (2004) Historical Aspects of EPR Parameter Calculations. In: Kaupp,
M.; B
uhl, M.; Malkin, V. (Eds) Calculation of NMR and EPR Parameters. Theory and Applications.
Wiley-VCH, pp 2132.
330. Neese, F. (2004) Zero-Field Splitting. In: Kaupp, M.; B

uhl, M.; Malkin, V. (Eds) Calculation of NMR
and EPR Parameters. Theory and Applications. Wiley-VCH, pp 541564.
331. Neese, F. (2004) Application of EPR Parameter Calculations in Bioinorganic Chemistry. In: Kaupp,
M.; B
uhl, M.; Malkin, V. (Eds) Calculation of NMR and EPR Parameters. Theory and Applications.
Wiley-VCH, pp 581591.
332. Neese, F. (2003) Quantum Chemical Calculations of Spectroscopic Properties of Metalloproteins and
Model Compounds: EPR and M ossbauer Properties. Curr. Op. Chem. Biol., 7, 125135.
333. Neese, F.; Solomon, E.I. (2003) Calculation and Interpretation of Spin-Hamiltonian Parameters in
Transition Metal Complexes. Invited review, (Wiley series: Magnetoscience From Molecules to
Materials edited by J.S. Miller and M. Drillon), Volume IV, p 345466.
783
Bibliography
[1] Grimme, S. (2006) J. Comput. Chem., 27, 1787.
[2] Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. (2010) J. Chem. Phys., 132, 154104.
[3] Zheng, J.; Xu, X.; Truhlar, D. G. (2010) Theor. Chem. Acc., 128, 295.
[4] Rappoport, D.; Furche, F. (2010) J. Chem. Phys., 133, 134105.
[5] Pantazis, D. A.; Chen, X.-Y.; Landis, C. R.; Neese, F. (2008) J. Chem. Theory Comput., 4, 908.
[6] Buhl, M.; Reimann, C.; Pantazis, D. A.; Bredow, T.; Neese, F. (2008) J. Chem. Theory Comput., 4,
1449.
[7] Pantazis, D. A.; Neese, F. (2009) J. Chem. Theory Comput., 5, 2229.
[8] Pantazis, D. A.; Neese, F. (2011) J. Chem. Theory Comput., 7, 677.
[9] Pantazis, D. A.; Neese, F. (2012) Theor. Chem. Acc., 131, 1292.
[10] Aravena, D.; Neese, F.; A., P. D. (2016) J. Chem. Theory Comput., 12, 1148.
[11] Hay, P. J.; Wadt, W. R. (1985) J. Chem. Phys., 82, 270 & 284 & 299.
[12] Fuentealba, P.; Preuss, H.; Stoll, H.; von Szentpaly, L. (1982) Chem. Phys. Lett., 89, 418.
[13] von Szentpaly, L.; Fuentealba, P.; Preuss, H.; Stoll, H. (1982) Chem. Phys. Lett., 93, 555.
[14] Fuentealba, P.; Stoll, H.; von Szentpaly, L.; Schwerdtfeger, P.; Preuss, H. (1983) J. Phys. B, 16, L323.
[15] Stoll, H.; P. Fuentealba, P. S.; Flad, J.; von Szentpaly, L.; Preuss, H. (1984) J. Chem. Phys., 81, 2732.
[16] Fuentealba, P.; von Szentpaly, L.; Preuss, H.; Stoll, H. (1985) J. Phys. B, 18, 1287.
[17] Dolg, M.; Wedig, U.; Stoll, H.; Preuss, H. (1987) J. Chem. Phys., 86, 866.
[18] Igel-Mann, G.; Stoll, H.; Preuss, H. (1988) Mol. Phys., 65, 1321.
[19] Dolg, M.; Stoll, H.; Preuss, H. (1989) J. Chem. Phys., 90, 1730.
[20] Schwerdtfeger, P.; Dolg, M.; Schwarz, W. H. E.; Bowmaker, G. A.; Boyd, P. D. W. (1989) J. Chem.
Phys., 91, 1762.
[21] Dolg, M.; Stoll, H.; Savin, A.; Preuss, H. (1989) Theor. Chim. Acta, 75, 173.
[22] Andrae, D.; H

auermann, U.; Dolg, M.; Stoll, H.; Preuss, H. (1990) Theor. Chim. Acta, 77, 123.
[23] Kaupp, M.; v. R. Schleyer, P.; Stoll, H.; Preuss, H. (1991) J. Chem. Phys., 94, 1360.
[24] K
uchle, W.; Dolg, M.; Stoll, H.; Preuss, H. (1991) Mol. Phys., 74, 1245.
[25] Dolg, M.; Fulde, P.; K

uchle, W.; Neumann, C.-S.; Stoll, H. (1991) J. Chem. Phys., 94, 3011.
[26] Dolg, M.; Stoll, H.; Flad, H.-J.; Preuss, H. (1992) J. Chem. Phys., 97, 1162.
[27] Bergner, A.; Dolg, M.; K

uchle, W.; Stoll, H.; Preuss, H. (1993) Mol. Phys., 80, 1431.
784 Bibliography
[28] Dolg, M.; Stoll, H.; Preuss, H.; Pitzer, R. M. (1993) J. Phys. Chem., 97, 5852.
[29] Dolg, M.; Stoll, H.; Preuss, H. (1993) Theor. Chim. Acta, 85, 441.
auermann, U.; Dolg, M.; Stoll, H.; Preuss, H. (1993) Mol. Phys., 78, 1211.
[30] H
[31] K
uchle, W.; Dolg, M.; Stoll, H.; Preuss, H. (1994) J. Chem. Phys., 100, 7535.
[32] Nicklass, A.; Dolg, M.; Stoll, H.; Preuss, H. (1995) J. Chem. Phys., 102, 8942.
[33] Leininger, T.; Nicklass, A.; Stoll, H.; Dolg, M.; Schwerdtfeger, P. (1996) J. Chem. Phys., 105, 1052.
[34] Leininger, T.; Nicklass, A.; K

uchle, W.; Stoll, H.; Dolg, M.; Bergner, A. (1996) Chem. Phys. Lett.,
255, 274.
[35] Leininger, T.; Berning, A.; Nicklass, A.; Stoll, H.; Werner, H.-J.; Flad, H.-J. (1997) Chem. Phys., 217,
19.
[36] Schautz, F.; Flad, H.-J.; Dolg, M. (1998) Theor. Chem. Acc., 99, 231.
[37] Wang, Y.; Dolg, M. (1998) Theor. Chem. Acc., 100, 124.
[38] Metz, B.; Stoll, H.; Dolg, M. (2000) J. Chem. Phys., 113, 2563.
[39] Metz, B.; Schweizer, M.; Stoll, H.; Dolg, M.; Liu, W. (2000) Theor. Chem. Acc., 104, 22.
[40] Martin, J. M. L.; Sundermann, A. (2001) J. Chem. Phys., 114, 3408.
[41] Cao, X.; Dolg, M. (2001) J. Chem. Phys., 115, 7348.
[42] Stoll, H.; Metz, B.; Dolg, M. (2002) J. Comput. Chem., 23, 767.
[43] Cao, X.; Dolg, M. (2002) J. Mol. Struct.: THEOCHEM, 581, 139.
[44] Peterson, K. A. (2003) J. Chem. Phys., 119, 11099.
[45] Peterson, K. A.; Figgen, D.; Goll, E.; Stoll, H.; Dolg, M. (2003) J. Chem. Phys., 119, 11113.
[46] Figgen, D.; Rauhut, G.; Dolg, M.; Stoll, H. (2005) Chem. Phys., 311, 227.
[47] Lim, I. S.; Schwerdtfeger, P.; Metz, B.; Stoll, H. (2005) J. Chem. Phys., 122, 104103.
[48] Peterson, K. A.; Puzzarini, C. (2005) Theor. Chem. Acc., 114, 283.
[49] Yang, J.; Dolg, M. (2005) Theor. Chem. Acc., 113, 212.
[50] Lim, I. S.; Stoll, H.; Schwerdtfeger, P. (2006) J. Chem. Phys., 124, 034107.
[51] Peterson, K. A.; Shepler, B. C.; Figgen, D.; Stoll, H. (2006) J. Phys. Chem. A, 110, 13877.
[52] Peterson, K. A.; Figgen, D.; Dolg, M.; Stoll, H. (2007) J. Chem. Phys., 126, 124101.
[53] Moritz, A.; Cao, X.; Dolg, M. (2007) Theor. Chem. Acc., 117 & 118, 473 & 845.
[54] Moritz, A.; Dolg, M. (2008) Theor. Chem. Acc., 121, 297.
[55] H
ulsen, M.; Weigand, A.; Dolg, M. (2009) Theor. Chem. Acc., 122, 23.
[56] Figgen, D.; Peterson, K. A.; Dolg, M.; Stoll, H. (2009) J. Chem. Phys., 130, 164108.
[57] Weigand, A.; Cao, X.; Yang, J.; Dolg, M. (2010) Theor. Chem. Acc., 126, 117.
[58] Flores-Moreno, R.; Alvares-Mendez, R. J.; Vela, A.; K

oster, A. M. (2006) J. Comput. Chem., 27, 1009.
Bibliography 785
[59] Szabo, A.; Ostlund, N. S. (1989) Modern Quantum Chemistry: Introduction to Advanced Electronic
Structure Theory. Dover Publications.
URL http://books.google.de/books?id=6mV9gYzEkgIC
[60] McWeeny, R. (1992) Methods of Molecular Quantum Mechanics. 2nd Edition. Academic Press.
[61] Cremer, D. (1998) In: v. R. Schleyer, P. (editor), Encyclopedia of Computational Chemistry, 1706,
John Wiley and Sons.
[62] Saebo, S.; Alml

of, J. (1989) Chem. Phys. Lett., 154, 83.
[63] Head-Gordon, M.; Pople, J. A. (1988) Chem. Phys. Lett., 153, 503.
[64] Lauderdale, W. J.; Stanton, J. F.; Gauss, J.; Watts, J. D.; Bartlett, R. J. (1991) Chem. Phys. Lett.,
187, 21.
[65] Knowles, P. J.; Andrews, J. S.; Amos, R. D.; Handy, N. C.; Pople, J. A. (1991) Chem. Phys. Lett.,
186, 130.
[66] Pople, J. A.; Binkley, J. S.; Seeger, R. (1976) Int. J. Quant. Chem. Symp., 10, 1.
[67] Krishnan, R.; Frisch, M. J.; Pople, J. A. (1980) J. Chem. Phys., 72, 4244.
[68] Handy, N. C.; Knowles, P. J.; Somasundram, K. (1985) Theor. Chem. Acc., 68, 87.
[69] Weigend, F.; H

aser, M.; Patzelt, H.; Ahlrichs, R. (1998) Chem. Phys. Lett., 294, 143.
[70] Weigend, F.; H

aser, M. (1997) Theor. Chem. Acc., 97, 331.
[71] Feyereisen, M.; Fitzerald, G.; Komornicki, A. (1993) Chem. Phys. Lett., 208, 359.
[72] Bernholdt, D. E.; Harrison, R. J. (1996) Chem. Phys. Lett., 250, 477.
[73] Grimme, S. (2003) J. Chem. Phys., 118, 9095.
[74] Neese, F.; Schwabe, T.; Kossmann, S.; Schirmer, B.; Grimme, S. (2009) J. Chem. Theory Comput., 5,
3060.
[75] Pinski, P.; Riplinger, C.; Valeev, E. F.; Neese, F. (2015) J. Chem. Phys., 143, 034108.
[76] Pavosevic, F.; Pinski, P.; Riplinger, C.; Neese, F.; Valeev, E. (2016) J. Chem. Phys., 144, 144109.
[77] Lee, T. J.; Taylor, P. R. (1989) Int. J. Quant. Chem. Symp., 23, 199.
[78] Wennmohs, F.; Neese, F. (2008) Chem. Phys., 343, 217.
[79] Chong, D. P.; Langhoff, S. R. (1986) J. Chem. Phys., 84, 5606.
[80] Ahlrichs, R.; Scharf, P.; Ehrhardt, C. (1985) J. Chem. Phys., 82, 890.
[81] Ahlrichs, R.; Scharf, P. (1987) In: Lawley, K. P. (editor), Advances in Chemical Physics: Ab Initio
Methods in Quantum Chemistry, Part I, Advances in Chemical Physics, Wiley.
[82] Neese, F.; Hansen, A.; Liakos, D. G. (2009) J. Chem. Phys., 131, 064103.
[83] Liakos, D. G.; Neese, F. (2012) J. Phys. Chem. A, 116, 4801.
ak, R.; Valeev, E. F.; Neese, F. (2013) Mol. Phys., 111, 2653.
[84] Liakos, D. G.; Izs
[85] Neese, F.; Hansen, A.; Wennmohs, F.; Grimme, S. (2009) Acc. Chem. Res., 42, 641.
[86] Neese, F.; Wennmohs, F.; Hansen, A. (2009) J. Chem. Phys., 130, 114108.
[87] Liakos, D. G.; Hansen, A.; Neese, F. (2011) J. Chem. Theory Comput., 7, 76.
786 Bibliography
[88] Hansen, A.; Liakos, D. G.; Neese, F. (2011) J. Chem. Phys., 135, 214102.
[89] Riplinger, C.; Neese, F. (2013) J. Chem. Phys., 138, 034106.
[90] Riplinger, C.; Sandhoefer, B.; Hansen, A.; Neese, F. (2013) J. Chem. Phys., 139, 134101.
[91] Riplinger, C.; Pinski, P.; Becker, U.; Valeev, E. F.; Neese, F. (2016) J. Chem. Phys., 144, 024109.
[92] Datta, D.; Kossmann, S.; Neese, F. (2016) J. Chem. Phys., 114101.
[93] Saitow, M.; Becker, U.; Riplinger, C.; Valeev, E. F.; Neese, F. (2016) J. Chem. Phys., under review.
[94] Pulay, P.; Saebo, S.; Meyer, W. (1984) J. Chem. Phys., 81, 1901.
[95] Sch
utz, M.; Werner, H. J. (2000) Chem. Phys. Lett., 318, 370.
[96] Sch
utz, M.; Werner, H. J. (2001) J. Chem. Phys., 114, 661.
[97] Liakos, D. G.; Sparta, M.; Kesharwani, M. K.; Martin, J. M. L.; Neese, F. (2015) J. Chem. Theory
Comput., 11, 1525.
[98] Li, S.; Ma, J.; Jiang, Y. (2002) J. Comput. Chem., 23, 237.
[99] Li, S.; Shen, J.; Li, W.; Jiang, Y. (2006) J. Chem. Phys., 125, 074109.
[100] Li, W.; Piecuch, P.; Gour, J.; Li, S. (2009) J. Chem. Phys., 125, 114109.
[101] Rolik, M., Z.; Kallay (2011) J. Chem. Phys., 135, 104111.
[102] Guo, Y.; Li, W.; Li, S. (2014) J. Phys. Chem. A, 118(39), 8996.
[103] F zek, J. (1985) Chem. Phys., 97, 251.

orner, W.; Ladik, J.; Otto, P.; C
[104] Neese, F.; Wennmohs, F.; Hansen, A.; Becker, U. (2009) Chem. Phys., 356, 98.
[105] Kossmann, S.; Neese, F. (2009) Chem. Phys. Lett., 481, 240.
[107] Schwabe, T.; Grimme, S. (2006) Phys. Chem. Chem. Phys., 8, 4398.
[108] Neese, F.; Schwabe, T.; Grimme, S. (2007) J. Chem. Phys., 126, 124115.
[109] Grimme, S.; Ehrlich, S.; Goerigk, L. (2011) J. Comput. Chem., 32, 1456.
[110] Grimme, S. (2004) J. Comput. Chem., 25, 1463.
[111] Iikura, H.; Tsuneda, T.; Yanai, T.; Hirao, K. (2001) J. Chem. Phys., 115, 3540.
[112] Yanai, T.; Tew, D. P.; Handy, N. C. (2004) Chem. Phys. Lett., 393, 51.
[113] Tawada, Y.; Tsuneda, T.; Yanagisawa, S.; Yanai, T.; Hirao, K. (2004) J. Chem. Phys., 120, 8425.
[114] Chai, J.-D.; Head-Gordon, M. (2008) J. Chem. Phys., 128, 084106.
[115] Lin, Y.-S.; Li, G.-D.; Mao, S.-P.; Chai, J.-D. (2013) J. Chem. Theory Comput., 9, 263.
[116] Angeli, C.; Cimiraglia, R.; Evangelisti, S.; Leininger, T.; Malrieu, J.-P. (2001) J. Chem. Phys., 114,
10252.
[117] Angeli, C.; Cimiraglia, R.; Malrieu, J.-P. (2001) Chem. Phys. Lett., 350, 297.
[118] Angeli, C.; Cimiraglia, R.; Malrieu, J.-P. (2002) J. Chem. Phys., 117, 9138.
Bibliography 787
[119] Havenith, R. W. A.; Taylor, P. R.; Angeli, C.; Cimiraglia, R.; Ruud, K. (2004) J. Chem. Phys., 120,
4619.
[120] Schapiro, I.; Sivalingam, K.; Neese, F. (2013) J. Chem. Theory Comput., 9(8), 3567.
[121] Guo, Y.; Sivalingam, K.; Valeev, E. F.; Neese, F. (2016) J. Chem. Phys., 144(9), 094111.
[122] van Lenthe, E.; Baerends, E. J.; Snijders, J. G. (1993) J. Chem. Phys., 99(6), 4597.
[123] van W
ullen, C. (1998) J. Chem. Phys., 109, 392.
[124] Schlegel, H. B. (1987) In: Lawley, K. P. (editor), Advances in Chemical Physics: Ab initio methods in
quantum chemistry, Part I, vol. 67, 249, John Wiley and Sons.
[125] Schlegel, H. B. (1995) In: Yarkony, D. R. (editor), Modern Electronic Structure Theory, 459, World
Scientific.
[126] Schlegel, H. B. (1998) In: v. R. Schleyer, P. (editor), Encyclopedia of Computational Chemistry, 1136,
[127] Eckert, F.; Pulay, P.; Werner, H. J. (1997) J. Comput. Chem., 12, 1473.
[128] Horn, H.; Wei, H.; H

aser, M.; Ehrig, M.; Ahlrichs, R. (1991) J. Comput. Chem., 12, 1058.
[129] Baker, J. (1986) J. Comput. Chem., 7, 385.
[130] Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. (2008) J. Chem. Theory Comput., 4, 435.
[131] Harvey, J. N.; Aschi, M.; Schwarz, H.; Koch, W. (1998) Theor. Chem. Acc., 99, 95.
[132] Li, H.; Jensen, J. H. (2002) Theor. Chem. Acc., 107, 211.
[133] Ribas-Arino, J.; Marx, D. (2012) Chem. Rev., 112(10), 5412.
[134] DeBeer-George, S.; Petrenko, T.; Neese, F. (2008) J. Phys. Chem. A, 112, 12963.
[135] DeBeer-George, S.; Petrenko, T.; Neese, F. (2008) Inorg. Chim. Acta, 361, 965.
[136] Sorensen, L. K.; Guo, M.; Lindh, R.; Lundberg, M. (2016) J. Chem. Theory Comput., 115, 174.
[137] Bernadotte, S.; Atkins, A. J.; Jacob, C. R. (2012) J. Chem. Phys., 137, 204106.
[138] Ray, K.; DeBeer-George, S.; Solomon, E. I.; Wieghardt, K.; Neese, F. (2007) Chem. Eur. J., 13, 2783.
[139] Seeger, R.; Pople, J. A. (1977) J. Chem. Phys., 66, 3045.
[140] Bauernschmitt, R.; Ahlrichs, R. (1996) J. Chem. Phys., 104, 9047.
[141] Datta, D.; Kong, L.; Nooijen, M. (2011) J. Chem. Phys., 134(21), 214116.
[142] Datta, D.; Nooijen, M. (2012) J. Chem. Phys., 137(20), 204107.
[143] Demel, O.; Datta, D.; Nooijen, M. (2013) J. Chem. Phys., 138(13), 134108.
[144] Nooijen, M.; Demel, O.; Datta, D.; Kong, L.; Shamasundar, K. R.; Lotrich, V.; Huntington, L. M.;
Neese, F. (2014) J. Chem. Phys., 140(8), 081102.
[145] Huntington, L. M. J.; Nooijen, M. (2015) J. Chem. Phys., 142, 194111.
[146] Huntington, L. M. J.; Nooijen, M. (2016) J. Chem. Theory Comput., 12, 114.
[147] Liu, Z.; Demel, O.; Nooijen, M. (2015) J. Mol. Spectrosc., 311, 54.
[148] Liu, Z.; Huntington, L. M. J.; Nooijen, M. (2015) Mol. Phys., 113, 2999.
788 Bibliography
[149] Neugebauer, J.; Reiher, M.; Kind, C.; Hess, B. A. (2002) J. Comput. Chem., 23, 895.
[150] Petrenko, T.; Sturhahn, W.; Neese, F. (2007) Hyperfine Interact., 175, 165.
[151] Petrenko, T.; DeBeer-George, S.; Aliaga-Alcalde, N.; Bill, E.; Mienert, B.; Xiao, Y.; Guo, Y.; Sturhahn,
W.; Cramer, S. P.; Wieghardt, K.; Neese, F. (2007) J. Am. Chem. Soc., 129, 11053.
[152] Grimme, S. (2012) Chem. Eur. J., 18, 9955.
[153] London, F. (1937) Phys. Radium, 8, 397.
[154] Ditchfield, R. (1972) J. Chem. Phys., 56, 5688.
nski, M.; Ruud, K. (1999) Chem. Rev., 99, 293.

[155] Helgaker, T.; Jaszu
[156] Gauss, J. (2000) In: Grotendorst, J. (editor), Modern Methods and Algorithms of Quantum Chemistry,
vol. 3, 541592, John von Neumann Institute for Computing, NIC Series, J ulich.
[157] Mason, J. (1993) Solid State Nucl. Magn. Res., 2, 285.
[158] Auer, A. A.; Gauss, J. (2003) J. Chem. Phys., 118, 10407.
[159] Flaig, D.; Maurer, M.; Hanni, M.; Braunger, K.; Kick, L.; Thubauville, M.; Ochsenfeld, C. (2014) J.
[160] Pantazis, D. A.; Orio, M.; Petrenko, T.; Zein, S.; Bill, E.; Lubitz, W.; Messinger, J.; Neese, F. (2009)
Chem. Eur. J., 15, 5108.
[161] Neese, F. (2006) J. Am. Chem. Soc., 128, 10213.
[162] Neese, F.; Solomon, E. I. (1998) Inorg. Chem., 37, 6568.
[163] Pederson, M. R.; Khanna, S. N. (1999) Phys. Rev. B, 60, 9566.
[164] Neese, F. (2007) J. Chem. Phys., 127, 164112.
[165] Sinnecker, S.; Neese, F. (2006) J. Phys. Chem. A, 110, 12267.
[166] Riplinger, C.; Kao, J. P. Y.; Rosen, G. M.; Kathirvelu, V.; Eaton, G. R.; Eaton, S. S.; Kutateladze, A.;
Neese, F. (2009) J. Am. Chem. Soc., 131, 10092.
[167] Neese, F. (2002) Inorg. Chim. Acta, 337C, 181.
[168] R
omelt, M.; Ye, S.; Neese, F. (2009) Inorg. Chem., 48, 784.
[169] Ginsberg, A. P. (1980) J. Am. Chem. Soc., 102, 111.
[170] Noodleman, L. (1981) J. Chem. Phys., 74, 5737.
[171] Noodleman, L.; Davidson, E. R. (1985) Chem. Phys., 109, 131.
[172] Bencini, A.; Gatteschi, D. (1980) J. Am. Chem. Soc., 108, 5763.
[173] Yamaguchi, K.; Takahara, Y.; Fueno, T. (1986) In: Smith, V. H. (editor), Applied Quantum Chemistry,
155, Wiley, Reidel, Dordrecht.
[174] Soda, T.; Kitagawa, Y.; Onishi, T.; Takano, Y.; Shigeta, Y.; Nagao, H.; Yoshioka, Y.; Yamaguchi, K.
(2000) Chem. Phys. Lett., 319, 223.
[175] Mitoraj, M. P.; Michalak, A.; Ziegler, T. (2009) J. Chem. Theory Comput., 5(4), 962.
[176] Schneider, W.; Bistoni, G.; Sparta., M.; Riplinger, C.; Saitow, M.; Auer, A.; Neese, F. (2016) J. Chem.
Theory Comput., 12(10), 4778.
Bibliography 789
[177] Bistoni, G.; Auer, A. A.; Neese, F. (2017) Chem. Eur. J., 23(4), 865.
[178] Koch, W.; Holthausen, M. C. (2000) A Chemists Guide to Density Functional Theory. Wiley-VCH.
[179] Parr, R. G. (1994) Density Functional Theory of Atoms and Molecules. International Series of Mono-
graphs on Chemistry, Oxford University Press.
[180] Zhang, Y.; Yang, W. (1998) Phys. Rev. Lett., 80, 890.
[181] Hammer, B.; Hansen, L. B.; Nrskov, J. K. (1999) Phys. Rev. B, 59, 7413.
[182] Perdew, J. P.; Wang, Y. (1986) Phys. Rev. B, 33, 8800.
[183] Perdew, J. P.; Yue, W. (1986) Phys. Rev. B, 40, 3399.
[184] Perdew, J. P.; Burke, K.; Ernzerhof, M. (1996) Phys. Rev. Lett., 77, 3865.
[185] Lee, K.; Murray, E. D.; Kong, L.; Lundqvist, B. I.; Langreth, D. C. (2010) Phys. Rev. B, 82, 081101.
[186] Zhao, Y.; Truhlar, D. G. (2006) J. Chem. Phys., 125, 194101.
[187] Perdew, J. P.; Ruzsinsky, A.; Csonka, G. I.; Constantin, L. A.; Sun, J. (2009) Phys. Rev. Lett., 103,
026403.
[188] Perdew, J. P.; Ruzsinsky, A.; Csonka, G. I.; Constantin, L. A.; Sun, J. (2011) Phys. Rev. Lett., 106,
179902.
[189] Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M. J. (1994) J. Phys. Chem., 98, 11623.
[190] Adamo, C.; Barone, V. (1999) J. Chem. Phys., 110, 6158.
[191] Zhao, Y.; Truhlar, D. G. (2008) Theor. Chem. Acc., 120, 215.
[192] Handy, N. C.; Cohen, A. J. (2001) Mol. Phys., 99, 403.
[193] Xu, X.; Goddard, W. A., III (2004) Proc. Nat. Acad. Sci., 101, 2673.
[194] Zhao, Y.; Truhlar, D. G. (2005) J. Phys. Chem. A, 109, 5656.
[195] Becke, A. D. (1993) J. Chem. Phys., 98, 1372.
[196] Karton, A.; Tarnopolsky, A.; Lam`ere, J.-F.; Schatz, G. C.; Martin, J. M. L. (2008) J. Phys. Chem. A,
112, 12868.
[197] Goerigk, L.; Grimme, S. (2011) J. Chem. Theory Comput., 7, 291.
[198] Dirac, P. A. M. (1930) Proc. Camb. Phil. Soc., 26, 376.
[199] Slater, J. C. (1974) The quantum theory of atoms molecules and solids, Vol. 4. McGraw Hill, New
York.
[200] Becke, A. D. (1988) Phys. Rev. A, 38, 3098.
[201] Gill, P. M. W. (1996) Mol. Phys., 89, 433.
[202] Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M. R.; Singh, D. J.; Fiolhais, C.
(1992) Phys. Rev. A, 46, 6671.
[203] Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M. R.; Singh, D. J.; Fiolhais, C.
(1993) Phys. Rev. A, 48, 4978.
[204] Adamo, C.; Barone, V. (1998) J. Chem. Phys., 108, 664.
[205] Hoe, W.-M.; Cohen, A. J.; Handy, N. C. (2001) Chem. Phys. Lett., 341, 319.
790 Bibliography
[206] Staroverov, V. N.; Scuseria, G. E.; Tao, J.; Perdew, J. P. (2003) J. Chem. Phys., 119, 12129.
[208] Vosko, S. H.; Wilk, L.; Nusair, M. (1980) Can. J. Phys., 58, 1200.
[209] Perdew, J. P.; Wang, Y. (1992) Phys. Rev. B, 45, 13244.
[210] Perdew, J. P. (1986) Phys. Rev. B, 33, 8822.
[211] Lee, C.; Yang, W.; Parr, R. G. (1988) Phys. Rev. B, 37, 785.
[213] Ahlrichs, R.; B

ar, M.; Baron, H. P.; Bauernschmitt, R.; B ocker, S.; Ehrig, M.; Eichkorn, K.; Elliott,
S.; Furche, F.; Haase, F.; Haser, M.; Horn, H.; Huber, C.; Huniar, U.; Kattanek, M.; Kolmel, C.;

Kollwitz, M.; May, K.; Ochsenfeld, C.; Ohm, H.; Sch
afer, A.; Schneider, U.; Treutler, O.; von Arnim,
M.; Weigend, F.; Weis, P.; Weiss, H. (2000) TurboMole - Program System for ab initio Electronic
Structure Calculations, Version 5.2. Universitat Karlsruhe, Karlsruhe, Germany.
[214] Ahlrichs, R. (1998) In: v. R. Schleyer, P. (editor), Encyclopedia of Computational Chemistry, 3123,
[215] Ahlrichs, R.; B

ar, M.; H
aser, M.; Horn, H.; Kolmel, C. (1989) Chem. Phys. Lett., 162, 165.
[216] Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.;
Zakrzewski, V. G.; Montgomery, J. A., Jr.; Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam,
J. M.; Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.; Petersson, G. A.; Ayala, P. Y.;
Cui, Q.; Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski,
J.; Ortiz, J. V.; Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.;
Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.;
Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Gonzalez, C.;
Head-Gordon, M.; Replogle, E. S.; Pople, J. A. (1998) Gaussian 98, Revision A.8. Gaussian, Inc.,
Pittsburgh PA.
[217] Hertwig, R. H.; Koch, W. (1997) Chem. Phys. Lett., 268, 345.
[218] Ernzerhof, M. (1998) In: Joubert, D. P. (editor), Density Functionals: Theory and Applications,
Springer Verlag.
[219] Adamo, C.; di Matteo, A.; Barone, V. (2000) Adv. Quant. Chem., 36, 45.
[221] Treutler, O.; Ahlrichs, R. J. (1994) J. Chem. Phys., 102, 346.
[222] Boerrigter, P. M.; Te Velde, G.; Baerends, E. J. (1988) Int. J. Quant. Chem., 33, 87.
[223] Baerends, E. J.; Ellis, D. E.; Ros, P. (1973) Chem. Phys., 2, 41.
[224] Dunlap, B. I.; Connolly, J. W. D.; Sabin, J. R. (1979) J. Chem. Phys., 71, 3396.
[225] Van Alsenoy, C. (1988) J. Comput. Chem., 9, 620.
[226] Kendall, R. A.; Fr

uchtl, H. A. (1997) Theor. Chem. Acc., 97, 158.

[227] Eichkorn, K.; Treutler, O.; Ohm, H.; Haser, M.; Ahlrichs, R. (1995) Chem. Phys. Lett., 240, 283.
[228] Eichkorn, K.; Weigend, F.; Treutler, O.; Ahlrichs, R. (1997) Theor. Chem. Acc., 97, 119.
[229] Whitten, J. L. (1973) J. Chem. Phys., 58, 4496.

Bibliography 791
[230] Te Velde, G.; Baerends, E. J. (1992) J. Comp. Phys., 99, 84.
[231] Murray, C. W.; Handy, N. C.; Laming, G. J. (1993) Mol. Phys., 78, 997.
[232] Stratmann, R. E.; Scuseria, G. E.; Frisch, M. J. (1996) Chem. Phys. Lett., 257, 213.
[233] Johnson, B. G.; Gill, P. M. W.; Pople, J. A. (1994) Chem. Phys. Lett., 220, 377.
[234] Gill, P. M. W.; Johnson, B. G.; Pople, J. A. (1993) Chem. Phys. Lett., 209, 506.
[235] Zheng, Y. C.; Alml

of, J. (1993) Chem. Phys. Lett., 214, 397.
[236] Zheng, Y. C.; Alml

of, J. (1996) J. Mol. Struct.: THEOCHEM, 388, 277.
[237] Berghold, G.; Hutter, J.; Parinello, M. (1998) Theor. Chem. Acc., 99, 344.
[238] Krack, M.; K

oster, A. M. (1998) J. Chem. Phys., 108, 3226.
[239] Lebedev, V. I. (1975) Zh. Vychisl. Mat. Fiz., 15, 48.
[240] Lebedev, V. I. (1976) Zh. Vychisl. Mat. Fiz., 16, 293.
[241] Lebedev, V. I.; Skorokhodov, A. L. (1992) Sov. Phys.-Dokl., 45, 587.
[242] Lebedev, V. I.; Laikov, D. N. (1999) Doklady Mathematics, 59, 477.
[243] Eden, M.; Levitt, M. H. (1998) J. Magn. Res., 132, 220.
[244] Stroud, A. H. (1971) Approximate Calculation of Multiple Integrals. Prentice-Hall, Englewood Cliffs.
[245] Laikov, D.; Van W ullen, C., Software. http://www.ccl.net/cca/software/SOURCES/FORTRAN/

Lebedev-Laikov-Grids/. These routines are based on C code provided by Dmitri Laikov
([email protected]). Christoph van W ullen ([email protected]) has made
the effort to obtain the code and convert it to Fortran77, and he obtained Dmitris kind permission to
make the routines publically available.
[246] Clementi, E.; Raimondi, D. (1963) IBM Res. Note, NJ-27.
[247] Vahtras, O.; Alml

of, J.; Feyereisen, M. W. (1993) Chem. Phys. Lett., 213, 514.
[248] Izsak, R.; Neese, F. (2011) J. Chem. Phys., 135, 144105.
[249] Becke, A. D.; Johnson, E. R. (2005) J. Chem. Phys., 122, 154101.
[250] Johnson, E. R.; Becke, A. D. (2005) J. Chem. Phys., 123, 024101.
[251] Johnson, E. R.; Becke, A. D. (2006) J. Chem. Phys., 124, 174104.
[252] Grimme, S., AK Grimme Homepage. http://www.thch.uni-bonn.de/tc/grimme.
[253] Vydrov, O. A.; Van Voorhis, T. (2010) J. Chem. Phys., 133, 244103.
[254] Hujo, W.; Grimme, S. (2011) J. Chem. Theory Comput.
[255] Goerigk, L.; Grimme, S. (2010) J. Chem. Theory Comput., 6, 107.
[256] Goerigk, L.; Grimme, S. (2011) Phys. Chem. Chem. Phys., 13, 6670.
ac, J.; Riley, K. E.; Hobza, P. (2011) J. Chem. Theory Comput., 7, 2427.
[257] Rez
[258] Hujo, W.; Grimme, S. (2011) Phys. Chem. Chem. Phys., 13, 13942.

[259] Jurecka, P.; Sponer,
J.; Cern
y, J.; Hobza, P. (2006) Phys. Chem. Chem. Phys., 8, 1985.
792 Bibliography
[260] Kruse, H.; Grimme, S. (2012) J. Chem. Phys., 16, 136.
[261] Kruse, H.; Goerigk, L.; Grimme, S. (2012) J. Org. Chem., 23, 10824.
[262] Sure, R.; Grimme, S. (2013) J. Comput. Chem., 34, 1672.
[263] Grimme, S.; Brandenburg, J. G.; Bannwarth, C.; Hansen, A. (2015) J. Chem. Phys., 143, 054107.
[264] Pople, J. A.; Beveridge, D. L. (1970) Approximate Molecular Orbital Theory. McGraw Hill Inc.
[265] Sedlej, J.; Cooper, I. L. (John Wiley and Sons) Semi-Emipirical Methods of Quantum Chemistry.
1985.
[266] Dewar, M. J. S.; Thiel, W. (1977) Theor. Chim. Acta, 46, 89.
[267] Thiel, W.; Voityuk, A. A. (1992) Theor. Chim. Acta, 81, 391.
[268] Dewar, M. J. S.; Thiel, W. (1977) J. Am. Chem. Soc., 99, 4899.
[269] Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. P. (1985) J. Am. Chem. Soc., 107, 3902.
[270] Stewart, J. P. (1989) J. Comput. Chem., 10, 209 & 221.
[271] Pople, J. A.; Segal, G. A. (1965) J. Chem. Phys., 43, 136.
[272] Pople, J. A.; Segal, G. A. (1966) J. Chem. Phys., 44, 3289.
[273] Santry, D. P. (1968) J. Am. Chem. Soc., 90, 3309.
[274] Santry, D. P.; Segal, G. A. (1967) J. Chem. Phys., 47, 158.
[275] Pople, J. A.; Beveridge, D. L.; Dobosh, P. A. (1967) J. Chem. Phys., 47, 2026.
[276] Clack, D. W.; Hush, N. S.; Yandle, J. R. (1972) J. Chem. Phys., 57, 3503.
[277] Clack, D. W. (1974) Mol. Phys., 27, 1513.
[278] Clack, D. W.; Smith, W. (1974) Theor. Chim. Acta, 36, 87.
[279] B
ohm, M. C.; Gleiter, R. (1981) Theor. Chim. Acta, 59, 127 & 153.
[280] Ridley, J.; Zerner, M. C. (1973) Theor. Chim. Acta, 32, 111.
[281] Bacon, A. D.; Zerner, M. C. (1979) Theor. Chim. Acta, 53, 21.
[282] Zerner, M. C.; Loew, G. H.; Kirchner, R. F.; Mueller-Westerhoff, U. T. (1980) J. Am. Chem. Soc.,
102, 589.
[283] Anderson, W. P.; Edwards, W. D.; Zerner, M. C. (1986) Inorg. Chem., 25, 2728.
[284] Anderson, W. P.; Cundari, T. R.; Drago, R. S.; Zerner, M. C. (1990) Inorg. Chem., 29, 3.
[285] Anderson, W. P.; Cundari, T. R.; Zerner, M. C. (1991) Int. J. Quant. Chem., XXXIX, 31.
[286] Zerner, M. C. (1990) In: Lipkowitz, K. B.; Boyd, D. B. (editors), Reviews in Computational Chemistry,
vol. 2, 313, Wiley-VCH.
[287] Zerner, M. C. (1992) In: Salahub, D. R.; Russo, N. (editors), Metal-Ligand Interactions: from Atoms
to Clusters to Surfaces, 101, Kluwer Academic Publishers.
[288] Zerner, M. C. (1992) In: Salahub, D. R.; Russo, N. (editors), Metal-Ligand Interactions: Structure
and Reactivity, 493, Kluwer Academic Publishers.
[289] Cory, M. G.; Zerner, M. C. (1991) Chem. Rev., 91, 813.

Bibliography 793
[290] Kotzian, M.; R

osch, N.; Zerner, M. C. (1992) Theor. Chim. Acta, 81, 201.
[291] Nieke, C.; Reinhold, J. (1984) Theor. Chim. Acta, 65, 99.
[292] K
ohler, H. J.; Birnstock, F. (1972) Z. Chem., 5, 196.
[293] Facility, M. S. C. (2000), Extensible Computational Chemistry Environment Basis Set Database. http:
//www.emsl.pnl.gov:2080/forms/basisform.html. Basis sets were obtained from the Extensible
Computational Chemistry Environment Basis Set Database, Version Mon Apr 17 10:05:30 PDT
2000, as developed and distributed by the Molecular Science Computing Facility, Environmental
and Molecular Sciences Laboratory which is part of the Pacific Northwest Laboratory, P.O. Box
999, Richland, Washington 99352, USA, and funded by the U.S. Department of Energy. The Pacific
Northwest Laboratory is a multi-program laboratory operated by Battelle Memorial Institue for the
U.S. Department of Energy under contract DE-AC06-76RLO 1830. Contact David Feller or Karen
Schuchardt for further information.
[294] Stoychev, G. L.; Auer, A. A.; Neese, F. (2017) J. Chem. Theory Comput., 13(2), 554.
[295] Mitin, A. V.; Hirsch, G.; Buenker, R. (1996) Chem. Phys. Lett., 259, 151.
[296] Mitin, A. V.; Hirsch, G.; Buenker, R. (1997) J. Comput. Chem., 18, 1200.
[297] Giordano, L.; Pacchioni, G.; Bredow, T.; Sanz, J. F. (2001) Surf. Sci., 471, 21.
[298] Amos, A. T.; Hall, G. G. (1961) Proc. R. Soc. Ser. A., 263, 483.
[299] King, H. F.; Stanton, R. E.; Kim, H.; Wyatt, R. E.; Parr, R. G. (1967) J. Chem. Phys., 47, 1936.
[300] Zerner, M. C.; Hehenberger, M. (1979) Chem. Phys. Lett., 62, 550.
[301] Guest, M. F.; Saunders, V. R. (1974) Mol. Phys., 28, 819.
[302] Saunders, V. R.; Hillier, I. H. (1973) Int. J. Quant. Chem., VII, 699.
[303] Pulay, P. (1980) Chem. Phys. Lett., 73, 393.
[304] Pulay, P. (1992) J. Comput. Chem., 3, 556.
[305] Fischer, T. H.; Alml

of, J. (1992) J. Phys. Chem., 96, 9768.
[306] Neese, F. (2000) Chem. Phys. Lett., 325, 93.
[307] Grimme, S.; Hansen, A. (2015) Angew. Chem. Int. Ed., 54, 12308.
[308] Dewar, M. J. S.; Hashmall, J. A.; Venier, C. G. (1968) J. Am. Chem. Soc., 90, 1953.
[309] McWeeny, R. (1974) Mol. Phys., 28, 1273.
[310] Brobowicz, F. W.; Goddard, W. A. (1977) In: III, H. F. S. (editor), Methods of Electronic Structure
Theory, 79, Plenum Press.
[311] Carbo, R.; Riera, J. M. (1978) A General SCF theory. Lecture notes in Chemistry. Springer Verlag.
[312] Binkley, J. S.; Pople, J. A.; Dobosh, P. A. (1974) Mol. Phys., 28, 1423.
[313] Edwards, W. D.; Zerner, M. C. (1987) Theor. Chim. Acta, 72, 347.
[314] Muller, R. P.; Langlois, J. M.; Ringnalda, M. N.; Friesner, R. A.; Goddard, W. A. (1994) J. Chem.
Phys., 100, 1226.
[315] Kollmar, C. (1996) J. Chem. Phys., 105, 8204.
[316] Kollmar, C. (1997) Int. J. Quant. Chem., 62, 617.

794 Bibliography
[317] Bofill, J. M.; Bono, H.; Rubio, J. (1998) J. Comput. Chem., 19, 368.
[318] Stavrev, K. K.; Zerner, M. C. (1997) Int. J. Quant. Chem., 65, 877.
[319] Zerner, M. C. (1989) Int. J. Quant. Chem., XXXV, 567.
[320] Alml
of, K., J. Faegri; Korsell, K. (1982) J. Comput. Chem., 3, 385.
[321] Almlof, J.; Taylor, P. R. (1984) In: Dykstra, C. E. (editor), Advanced Theories and Computational
Approaches to the Electronic Structure of Molecules, 107, Springer.
[322] Almlof, J. (1995) In: Yarkony, D. R. (editor), Modern Electronic Structure Theory, 110, World
Scientific.
aser, M.; Ahlrichs, R. (1989) J. Comput. Chem., 10, 104.

[323] H
[324] Jensen, F. (1999) Introduction to Computational Chemistry. Wiley.
[325] Ahlrichs, R. (1979) Comp. Phys. Comm., 17, 31.
[326] Gdanitz, R. J. (2001) Int. J. Quant. Chem., 85, 281.
[327] Gdanitz, R. J.; Ahlrichs, R. (143) Chem. Phys. Lett., 1988, 413.
[328] Szalay, P. G.; Bartlett, R. J. (1993) Chem. Phys. Lett., 214, 481.
[329] Kollmar, C.; Neese, F. (2010) Mol. Phys., 108, 2449.
[330] Kollmar, C.; Neese, F. (2011) J. Chem. Phys., 135, 064103.
[331] Scuseria, G. E.; III, H. F. S. (1987) Chem. Phys. Lett., 142, 354.
[332] Handy, N. C.; Pople, J. A.; Head-Gordon, M.; Raghavachari, K.; Trucks, G. W. (1989) Chem. Phys.
Lett., 164, 185.
[333] Kollmar, C.; Hesselmann, A. (2010) Theor. Chem. Acc., 127, 311.
[334] Salter, E. A.; Trucks, G. W.; Bartlett, R. J. (1989) J. Chem. Phys., 90, 1752.
[335] Kollmar, C.; Neese, F. (2011) J. Chem. Phys., 135, 084102.
[336] Huntington, L. M. J.; Nooijen, M. (2010) J. Chem. Phys., 133, 184109.
[337] Huntington, L. M. J.; Hansen, A.; Neese, F.; Nooijen, M. (2012) J. Chem. Phys., 136, 064101.
[338] Hampel, C.; Peterson, K. A.; Werner, H. J. (1992) Chem. Phys. Lett., 190, 1.
[339] Scuseria, G. E.; Janssen, C. L.; III, H. F. S. (1988) J. Chem. Phys., 89, 7382.
[340] Heully, J. L.; Malrieu, J.-P. (2006) J. Mol. Struct.: THEOCHEM, 768, 53.
[341] Neese, F.; Liakos, D. G.; Ye, S. F. (2011) J. Biol. Inorg. Chem., 16, 821.
[342] Jeziorski, B.; Monkhorst, H. J. (1981) Phys. Rev. A, 24, 1668.
[343] Mahapatra, U. S.; Datta, B.; Mukherjee, D. (1999) J. Chem. Phys., 110, 6171.
[344] M
asik, J.; Hubac, I. (1998) Adv. Quant. Chem., 31, 75.

[345] Bhaskaran-Nair, K.; Demel, O.; Smydke, J.; Pittner, J. (2011) J. Chem. Phys., 134, 154106.
[346] Demel, O.; Pittner, J. (2006) J. Chem. Phys., 124, 144112.
[347] Dyall, K. G. (1995) J. Chem. Phys., 102, 4909.

Bibliography 795
[348] Atanasov, M.; Ganyushin, D.; Sivalingam, K.; Neese, F. (2011) In: Mingos, D. M. P.; Day, P.; Dahl,
J. P. (editors), Molecular Electronic Structures of Transition Metal Complexes II, no. 143 in Structure
and Bonding, 149220, Springer Berlin Heidelberg.
[349] Atanasov, M.; Ganyushin, D.; Pantazis, D. A.; Sivalingam, K.; Neese, F. (2011) Inorg. Chem., 50(16),
7460.
[350] Aravena, D.; Atanasov, M.; Neese, F. (2016) Inorg. Chem., 55(9), 4457.
[351] Chan, G. K.-L.; Head-Gordon, M. (2002) J. Chem. Phys., 116, 4462.
[352] Chan, G. K.-L. (2004) J. Chem. Phys., 120, 3172.
[353] Ghosh, D.; Hachmann, J.; Yanai, T.; Chan, G. K.-L. (2008) J. Chem. Phys., 128, 144117.
[354] Sharma, S.; Chan, G. K.-L. (2012) J. Chem. Phys., 136, 124121.
[355] Ganyushin, D.; Neese, F. (2013) J. Chem. Phys., 138, 104113.
[356] Guo, S.; Watson, M. A.; Hu, W.; Sun, Q.; Chan, G. K.-L. (2016) J. Chem. Theory Comput., 12(4),
1583.
URL http://dx.doi.org/10.1021/acs.jctc.5b01225
[357] Guo, Y.; Sivalingam, K.; Valeev, E. F.; Neese, F. (to be submitted) J. Chem. Phys.
[358] Domingo, A.; Carvajal, M.-A.; de Graaf, C.; Sivalingam, K.; Neese, F.; Angeli, C. (2012) Theor. Chem.
Acc., 131(9), 1264.
[359] Angeli, C.; Borini, S.; Cestari, M.; Cimigraglia, R. (2004) J. Chem. Phys., 121, 4043.
[360] des Cloizeaux, J. (1960) Nucl. Phys., 20, 321.
[361] Chan, G. K.-L.; Sharma, S. (2011) Ann. Rev. Phys. Chem., 62, 465.
[362] Chan, G. K.-L., DMRG Homepage. http://www.princeton.edu/chemistry/chan/software/dmrg/.
[363] Fiedler, M. (1973) Czech. Math. J., 23, 298.
[364] Fiedler, M. (1975) Czech. Math. J., 25, 619.
[365] Atkins, J. E.; Boman, E. G.; Hendrickson, B. (1998) SIAM J. Comput., 28(1), 297.
[366] Barcza, G.; Legeza, O.; Marti, K. H.; Reiher, M. (2011) Phys. Rev. A, 83, 012508.
[367] van Lenthe, E.; Baerends, E. J.; Snijders, J. G. (1994) J. Chem. Phys., 101, 9783.
[368] Sandhoefer, B.; Neese, F. (2012) J. Chem. Phys., 137, 094102.
[369] Visscher, L.; Dyall, K. G. (1997) Atom. Data Nucl. Data Tabl., 67, 207.
[370] Siegbahn, P. E. M. (1980) Int. J. Quant. Chem., 18(5), 1229.
[371] Meyer, W. (1977) In: Schaefer III, H. F. (editor), Methods of Electronic Structure Theory, 413446,
Springer US.
[372] Sivalingam, K.; Krupicka, M.; Auer, A. A.; Neese, F. (2016) J. Chem. Phys., 145(5), 054104.
[373] Neese, F. (2006) J. Biol. Inorg. Chem., 11, 702.
[374] Neese, F. (2009) Coord. Chem. Rev., 253, 526.
[375] Petrenko, T.; Kossmann, S.; Neese, F. (2011) J. Chem. Phys., 134, 054116.

796 Bibliography
[377] Bannwarth, C.; Grimme, S. (2014) Comp. Theor. Chem., 10401041, 45.
[378] Risthaus, T.; Hansen, A.; Grimme, S. (2014) Phys. Chem. Chem. Phys., 16, 14408.
[379] Head-Gordon, M.; Rico, R. A.; Oumi, M.; Lee, T. J. (1994) Chem. Phys. Lett., 219, 21.
[380] Gr
uning, M.; Gritsenko, O. V.; Gisbergen, S. J. A.; Baerends, E. J. (2001) J. Chem. Phys., 114, 652.
[381] Roemelt, M.; Neese, F. (2013) J. Phys. Chem. A, 117, 3069.
[382] Plasser, F.; Wormit, M.; Dreuw, A. (2014) J. Chem. Phys., 141, 024106.
[383] Dutta, A. K.; Neese, F.; Izs

ak, R. (2016) J. Chem. Phys., 144(3), 034102.
[384] Dutta, A. K.; Nooijen, M.; Neese, F.; Izsak, R. (2017) J. Chem. Phys., 146(7).
ak, R. (2016) J. Chem. Phys., 145(3), 034102.

[385] Dutta, A. K.; Neese, F.; Izs
[386] Neese, F.; Petrenko, T.; Ganyushin, D.; Olbrich, G. (2007) Coord. Chem. Rev., 251(3-4), 288.
[387] Retegan, M.; Cox, N.; Pantazis, D. A.; Neese, F. (2014) Inorg. Chem., 53(21), 11785.
[388] Neese, F. (2003) Chem. Phys. Lett., 380, 721.
[389] Maganas, D.; Sottini, S.; Kyritsis, P.; Groenen, E. J. J.; Neese, F. (2011) Inorg. Chem., 50, 8741.
[390] Jiang, S.; Maganas, D.; Levesanos, N.; Ferentinos, E.; Haas, S.; Thirunavukkuarasu, K.; Krzystek, J.;
Dressel, M.; Bogani, L.; Neese, F.; Kyritsis, P. (2015) J. Am. Chem. Soc., 137, 12923.
[391] Ganyushin, D.; Neese, F. (2008) J. Chem. Phys., 128, 114117.
[392] Mukherjee, D. (1997) Chem. Phys. Lett., 274, 561.
[393] Kutzelnigg, W.; Mukherjee, D. (1997) J. Chem. Phys., 107, 432.
[394] National Institute of Standards and Technology (NIST) Atomic Spectra Database.
[395] Nave, G.; Johansson, S.; Learner, R. C. M.; Thorne, A. P.; Brault, J. W. (1994) Astrophys. J., Suppl.
Ser., 94, 221.
[396] Barone, V.; Cossi, M. (1998) J. Phys. Chem. A, 102, 1995.
[397] Marenich, A. V.; Cramer, C. J.; Truhlar, D. G. (2009) J. Phys. Chem. B, 113, 6378.
[398] Pascual-Ahuir, J. L.; Silla, E. (1990) J. Comput. Chem., 11, 1047.
[399] Pascual-Ahuir, J. L.; Silla, E.; Tunon, I. (1991) J. Comput. Chem., 12, 1077.
[400] Pascual-Ahuir, J. L.; Silla, E.; Tunon, I. (1994) J. Comput. Chem., 15, 1127.
[401] Truong, T. N.; Stefanovich, E. V. (1995) Chem. Phys. Lett., 240, 253.
[402] Marenich, A.; Hawkins, G.; Liotard, D.; Cramer, D., GESOL - version 2008. https://comp.chem.umn.
edu/gesol.
[403] Neese, F. (2005) J. Chem. Phys., 122, 034107.
[404] Suturina, E. A.; Maganas, D.; Bill, E.; Atanasov, M.; Neese, F. (2015) Inorg. Chem., 54, 9948.
[405] Koseki, S.; Schmidt, M. W.; Gordon, M. S. (1992) J. Phys. Chem., 96, 10768.
[406] Koseki, S.; Gordon, M. S.; Schmidt, M. W.; Matsunaga, N. (1995) J. Phys. Chem., 99, 12764.
[407] Koseki, S.; Schmidt, M. W.; Gordon, M. S. (1998) J. Phys. Chem. A, 102, 10430.
Bibliography 797
[408] Harriman, J. E. (1978) Theoretical Foundations of Electron Spin Resonance: Physical Chemistr: A
series of monographs. Academic Press.
[409] Hess, B. A.; Marian, C. M.; Wahlgren, U.; Gropen, O. (1996) Chem. Phys. Lett., 251, 365.
[410] Schimmelpfennig, B. (1996), AMFI - an atomic mean-field spin-orbit integral program.
[411] Berning, A.; Schweizer, M.; Werner, H.-J.; Knowles, P. J.; Palmieri, P. (2000) Mol. Phys., 98, 1823.
[412] Mulliken, R. S. (1955) J. Chem. Phys., 23, 1833.
[413] Wiberg, K. B. (1968) Tetrahedron, 24, 1083.
[414] Mayer, I. (1983) Chem. Phys. Lett., 97, 270.
[415] Mayer, I. (1984) Int. J. Quant. Chem., 26, 151.
[416] Mayer, I. (1985) Theor. Chim. Acta, 67, 315.
[417] Mayer, I. (1987) In: Maksic, Z. B. (editor), Modelling of Structure and Properties of Molecules, John
Wiley and Sons.
[418] Takatsuka, K.; Fueno, T.; Yamaguchi, K. (1978) Theor. Chim. Acta, 48, 175.
[419] Head-Gordon, M. (2003) Chem. Phys. Lett., 372, 508.
[420] Breneman, C. M.; Wiberg, K. B. (1990) J. Comput. Chem., 11, 361.
[421] Knizia, G. (2013) J. Chem. Theory Comput., 9, 4834.

Orca

Uploaded by

Copyright:

Available Formats

Orca

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Orca

Uploaded by

Copyright:

Available Formats

ORC A

- An ab initio, DFT and semiempirical SCF-MO package -

Design and Scientific Directorship:

With contributions from:

And contributions from our collaborators:

Initial conversion from the original Word document to LATEX:

1 Foreword 2017 (Version 4.0) XVI

2 ORCA 4.0 Changes XIX

3 FAQ frequently asked questions XXIII

5 Calling the Program (Serial and Parallel) 3

6 General Structure of the Input File 9

8 Running Typical Calculations 40

8.1.10.2 ZORA and IORA . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133

8.10.10 Size Consistency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210

9 Detailed Documentation 277

9.2.2.9 The Standard Computational Levels . . . . . . . . . . . . . . . . . . 305

9.9.5 RIJCOSX-RI-MP2 Gradients . . . . . . . . . . . . . . . . . . . . . . . . . . . 374

9.16.2 Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474

9.21.3 The reliabilty of the calculated excitation energy . . . . . . . . . . . . . . . . 555

9.25.3 Fitting of Experimental Spectra . . . . . . . . . . . . . . . . . . . . . . . . . 634

9.31 Orbital and Density Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 730

10 Some Tips and Tricks 749

11 Publications Related to ORCA 756

9.1 Graphical description of the Range-Separation ansatz. . . . . . . . . . . . . . . . . . 290

9.4 Overview of parametrized basis sets. . . . . . . . . . . . . . . . . . . . . . . . . . . . 316

1 Foreword 2017 (Version 4.0)

A final word to the academic users of this program package:

Cite our original research and development papers

in your own work!

This will provide us with the possibility of raising sufficient funding

2 ORCA 4.0 Changes

2.1 New Features

Linear scaling DLPNO-CCSD(T) open shell. New restricted open-shell formulation

SCF, DFT and Hessian

Improved performance with RI-J with conventionally stored integrals

Automatic closed shell STEOM-CCSD active space selection

CASSCF and NEVPT2

TD-DFT and ROCIS

Full TD-DFT for hybrid functionals

ROCIS MCD spectra

New correlation consistent basis sets added

2.2 Changes of Default Settings & Parameters

2.3 Important Bugfixes with respect to ORCA 3.0.3

Batching in the RI-MP2 Hessian(wrong numbers)

3 FAQ frequently asked questions

Noticeable changes between ORCA 2.x, 3.x and ORCA 4.0

ORCA 4.0 exhibits a new basis sets convention.

From ORCA 4.0 onwards, the standard basis-sets contain ECPs

Old (pre 4.0) New (> ORCA 4.0)

%basis basis CC_PVDZ %basis basis "CC-PVDZ"

Why is ORCA called ORCA?

How do I install ORCA on Linux / MacOS / Windows?

Ive installed ORCA, how do I start it?

<full orca binary folder path>/orca example.inp > example.out

<full orca binary folder path>/orca example.inp > example.out

How do I cite ORCA?

The generic reference for ORCA is:

Are there recommended programmes to use alongside ORCA?

OpenBabel is very useful for file conversion to various chemical formats.

Exact J + exact K: no auxiliary functions and no grids needed.